fix small typo

* Remove explicity state machine code for all but replication.Replication
* Introduce explicit error types that satisfy interfaces which provide
  sufficient information for replication.Replication to make intelligent
  retry + queuing decisions

  * Temporary()
  * LocalToFS()

* Remove the queue and replace it with a simple array that we sort each
  time (yay no generics :( )

.gitignore

@@ -1,3 +1,13 @@
+# Build
+artifacts/
+
+# Golang
+vendor/
+
+# IDEs
+.idea/
+.vscode/
+
 # Compiled Object files, Static and Dynamic libs (Shared Objects)
 *.o
 *.a

.gitmodules

@@ -1,4 +0,0 @@
-[submodule "docs/themes/docdock"]
-	path = docs/themes/docdock
-	url = git@github.com:zrepl/hugo-theme-docdock.git
-

Gopkg.lock

@@ -0,0 +1,313 @@
+# This file is autogenerated, do not edit; changes may be undone by the next 'dep ensure'.
+
+
+[[projects]]
+  branch = "master"
+  digest = "1:c0bec5f9b98d0bc872ff5e834fac186b807b656683bd29cb82fb207a1513fabb"
+  name = "github.com/beorn7/perks"
+  packages = ["quantile"]
+  pruneopts = ""
+  revision = "3a771d992973f24aa725d07868b467d1ddfceafb"
+
+[[projects]]
+  digest = "1:56c130d885a4aacae1dd9c7b71cfe39912c7ebc1ff7d2b46083c8812996dc43b"
+  name = "github.com/davecgh/go-spew"
+  packages = ["spew"]
+  pruneopts = ""
+  revision = "346938d642f2ec3594ed81d874461961cd0faa76"
+  version = "v1.1.0"
+
+[[projects]]
+  digest = "1:e988ed0ca0d81f4d28772760c02ee95084961311291bdfefc1b04617c178b722"
+  name = "github.com/fatih/color"
+  packages = ["."]
+  pruneopts = ""
+  revision = "5b77d2a35fb0ede96d138fc9a99f5c9b6aef11b4"
+  version = "v1.7.0"
+
+[[projects]]
+  branch = "master"
+  digest = "1:5d0a2385edf4ba44f3b7b76bc0436ceb8f62bf55aa5d540a9eb9ec6c58d86809"
+  name = "github.com/ftrvxmtrx/fd"
+  packages = ["."]
+  pruneopts = ""
+  revision = "c6d800382fff6dc1412f34269f71b7f83bd059ad"
+
+[[projects]]
+  branch = "master"
+  digest = "1:4416ad04caf950d004921268f777e418f8c328080ccf6a91e799f977dc128808"
+  name = "github.com/gdamore/encoding"
+  packages = ["."]
+  pruneopts = ""
+  revision = "b23993cbb6353f0e6aa98d0ee318a34728f628b9"
+
+[[projects]]
+  digest = "1:d8088c614a5bffaf38a40c116226e7d230750bafb77942dca25eab779ceeaf1f"
+  name = "github.com/gdamore/tcell"
+  packages = [
+    ".",
+    "termbox",
+    "terminfo",
+  ]
+  pruneopts = ""
+  revision = "de7e78efa4a71b3f36c7154989c529dbdf9ae623"
+  version = "v1.1.0"
+
+[[projects]]
+  digest = "1:6a4a01d58b227c4b6b11111b9f172ec5c17682b82724e58e6daf3f19f4faccd8"
+  name = "github.com/go-logfmt/logfmt"
+  packages = ["."]
+  pruneopts = ""
+  revision = "390ab7935ee28ec6b286364bba9b4dd6410cb3d5"
+  version = "v0.3.0"
+
+[[projects]]
+  digest = "1:3dd078fda7500c341bc26cfbc6c6a34614f295a2457149fc1045cab767cbcf18"
+  name = "github.com/golang/protobuf"
+  packages = ["proto"]
+  pruneopts = ""
+  revision = "aa810b61a9c79d51363740d207bb46cf8e620ed5"
+  version = "v1.2.0"
+
+[[projects]]
+  branch = "master"
+  digest = "1:cb09475f771b9167fb9333629f5d6a7161572602ea040f1094602b0dc8709878"
+  name = "github.com/jinzhu/copier"
+  packages = ["."]
+  pruneopts = ""
+  revision = "db4671f3a9b8df855e993f7c94ec5ef1ffb0a23b"
+
+[[projects]]
+  branch = "master"
+  digest = "1:1ed9eeebdf24aadfbca57eb50e6455bd1d2474525e0f0d4454de8c8e9bc7ee9a"
+  name = "github.com/kr/logfmt"
+  packages = ["."]
+  pruneopts = ""
+  revision = "b84e30acd515aadc4b783ad4ff83aff3299bdfe0"
+
+[[projects]]
+  branch = "master"
+  digest = "1:591a2778aa6e896980757ea87e659b3aa13d8c0e790310614028463a31c0998b"
+  name = "github.com/kr/pretty"
+  packages = ["."]
+  pruneopts = ""
+  revision = "cfb55aafdaf3ec08f0db22699ab822c50091b1c4"
+
+[[projects]]
+  branch = "master"
+  digest = "1:109e0a9b9e74c5c8adf94a2dc4dd4a9ca2a183d4d87ffecd4e62db69a5ede55a"
+  name = "github.com/kr/text"
+  packages = ["."]
+  pruneopts = ""
+  revision = "7cafcd837844e784b526369c9bce262804aebc60"
+
+[[projects]]
+  digest = "1:863ba095ae0954100963826333bd1503c169424bb798d9ce800d9f6651f4c234"
+  name = "github.com/lucasb-eyer/go-colorful"
+  packages = ["."]
+  pruneopts = ""
+  revision = "345fbb3dbcdb252d9985ee899a84963c0fa24c82"
+  version = "v1.0"
+
+[[projects]]
+  digest = "1:9ea83adf8e96d6304f394d40436f2eb44c1dc3250d223b74088cc253a6cd0a1c"
+  name = "github.com/mattn/go-colorable"
+  packages = ["."]
+  pruneopts = ""
+  revision = "167de6bfdfba052fa6b2d3664c8f5272e23c9072"
+  version = "v0.0.9"
+
+[[projects]]
+  digest = "1:78229b46ddb7434f881390029bd1af7661294af31f6802e0e1bedaad4ab0af3c"
+  name = "github.com/mattn/go-isatty"
+  packages = ["."]
+  pruneopts = ""
+  revision = "0360b2af4f38e8d38c7fce2a9f4e702702d73a39"
+  version = "v0.0.3"
+
+[[projects]]
+  digest = "1:82b912465c1da0668582a7d1117339c278e786c2536b3c3623029a0c7141c2d0"
+  name = "github.com/mattn/go-runewidth"
+  packages = ["."]
+  pruneopts = ""
+  revision = "ce7b0b5c7b45a81508558cd1dba6bb1e4ddb51bb"
+  version = "v0.0.3"
+
+[[projects]]
+  digest = "1:4c23ced97a470b17d9ffd788310502a077b9c1f60221a85563e49696276b4147"
+  name = "github.com/matttproud/golang_protobuf_extensions"
+  packages = ["pbutil"]
+  pruneopts = ""
+  revision = "3247c84500bff8d9fb6d579d800f20b3e091582c"
+  version = "v1.0.0"
+
+[[projects]]
+  digest = "1:7365acd48986e205ccb8652cc746f09c8b7876030d53710ea6ef7d0bd0dcd7ca"
+  name = "github.com/pkg/errors"
+  packages = ["."]
+  pruneopts = ""
+  revision = "645ef00459ed84a119197bfb8d8205042c6df63d"
+  version = "v0.8.0"
+
+[[projects]]
+  digest = "1:256484dbbcd271f9ecebc6795b2df8cad4c458dd0f5fd82a8c2fa0c29f233411"
+  name = "github.com/pmezard/go-difflib"
+  packages = ["difflib"]
+  pruneopts = ""
+  revision = "792786c7400a136282c1664665ae0a8db921c6c2"
+  version = "v1.0.0"
+
+[[projects]]
+  branch = "master"
+  digest = "1:1392748e290ca66ac8447ef24961f8ae9e1d846a53af0f58a5a0256982ce0577"
+  name = "github.com/problame/go-netssh"
+  packages = ["."]
+  pruneopts = ""
+  revision = "c56ad38d2c91397ad3c8dd9443d7448e328a9e9e"
+
+[[projects]]
+  branch = "master"
+  digest = "1:8c63c44f018bd52b03ebad65c9df26aabbc6793138e421df1c8c84c285a45bc6"
+  name = "github.com/problame/go-rwccmd"
+  packages = ["."]
+  pruneopts = ""
+  revision = "391d2c78c8404a9683d79f75dd24ab53040f89f7"
+
+[[projects]]
+  digest = "1:1bcbb0a7ad8d3392d446eb583ae5415ff987838a8f7331a36877789be20667e6"
+  name = "github.com/problame/go-streamrpc"
+  packages = [
+    ".",
+    "internal/pdu",
+  ]
+  pruneopts = ""
+  revision = "d5d111e014342fe1c37f0b71cc37ec5f2afdfd13"
+  version = "v0.5"
+
+[[projects]]
+  branch = "master"
+  digest = "1:ebf8ffdde9bdbf9c83e22121875c68c01d821776523546554b2a3ff6f72773ab"
+  name = "github.com/prometheus/client_golang"
+  packages = [
+    "prometheus",
+    "prometheus/promhttp",
+  ]
+  pruneopts = ""
+  revision = "e11c6ff8170beca9d5fd8b938e71165eeec53ac6"
+
+[[projects]]
+  branch = "master"
+  digest = "1:60aca47f4eeeb972f1b9da7e7db51dee15ff6c59f7b401c1588b8e6771ba15ef"
+  name = "github.com/prometheus/client_model"
+  packages = ["go"]
+  pruneopts = ""
+  revision = "99fa1f4be8e564e8a6b613da7fa6f46c9edafc6c"
+
+[[projects]]
+  branch = "master"
+  digest = "1:af21ee3e0a8212f17bb317cd7237f9920bcb2641a291ac111f30f63b3cab817f"
+  name = "github.com/prometheus/common"
+  packages = [
+    "expfmt",
+    "internal/bitbucket.org/ww/goautoneg",
+    "model",
+  ]
+  pruneopts = ""
+  revision = "d0f7cd64bda49e08b22ae8a730aa57aa0db125d6"
+
+[[projects]]
+  branch = "master"
+  digest = "1:61df0898746840afc7be5dc2c3eeec83022fab70df11ecee5b16c85e912cf5ed"
+  name = "github.com/prometheus/procfs"
+  packages = [
+    ".",
+    "internal/util",
+    "nfs",
+    "xfs",
+  ]
+  pruneopts = ""
+  revision = "8b1c2da0d56deffdbb9e48d4414b4e674bd8083e"
+
+[[projects]]
+  branch = "master"
+  digest = "1:146327ce93be37e68bd3ff8541090d96da8cb3adc9e35d57570e9170a29f6bf6"
+  name = "github.com/spf13/cobra"
+  packages = ["."]
+  pruneopts = ""
+  revision = "b78744579491c1ceeaaa3b40205e56b0591b93a3"
+
+[[projects]]
+  digest = "1:261bc565833ef4f02121450d74eb88d5ae4bd74bfe5d0e862cddb8550ec35000"
+  name = "github.com/spf13/pflag"
+  packages = ["."]
+  pruneopts = ""
+  revision = "e57e3eeb33f795204c1ca35f56c44f83227c6e66"
+  version = "v1.0.0"
+
+[[projects]]
+  digest = "1:3926a4ec9a4ff1a072458451aa2d9b98acd059a45b38f7335d31e06c3d6a0159"
+  name = "github.com/stretchr/testify"
+  packages = [
+    "assert",
+    "require",
+  ]
+  pruneopts = ""
+  revision = "69483b4bd14f5845b5a1e55bca19e954e827f1d0"
+  version = "v1.1.4"
+
+[[projects]]
+  branch = "v2"
+  digest = "1:9d92186f609a73744232323416ddafd56fae67cb552162cc190ab903e36900dd"
+  name = "github.com/zrepl/yaml-config"
+  packages = ["."]
+  pruneopts = ""
+  revision = "af27d27978ad95808723a62d87557d63c3ff0605"
+
+[[projects]]
+  branch = "master"
+  digest = "1:9c286cf11d0ca56368185bada5dd6d97b6be4648fc26c354fcba8df7293718f7"
+  name = "golang.org/x/sys"
+  packages = ["unix"]
+  pruneopts = ""
+  revision = "bf42f188b9bc6f2cf5b8ee5a912ef1aedd0eba4c"
+
+[[projects]]
+  digest = "1:5acd3512b047305d49e8763eef7ba423901e85d5dd2fd1e71778a0ea8de10bd4"
+  name = "golang.org/x/text"
+  packages = [
+    "encoding",
+    "encoding/internal/identifier",
+    "internal/gen",
+    "transform",
+    "unicode/cldr",
+  ]
+  pruneopts = ""
+  revision = "f21a4dfb5e38f5895301dc265a8def02365cc3d0"
+  version = "v0.3.0"
+
+[solve-meta]
+  analyzer-name = "dep"
+  analyzer-version = 1
+  input-imports = [
+    "github.com/fatih/color",
+    "github.com/gdamore/tcell/termbox",
+    "github.com/go-logfmt/logfmt",
+    "github.com/golang/protobuf/proto",
+    "github.com/jinzhu/copier",
+    "github.com/kr/pretty",
+    "github.com/mattn/go-isatty",
+    "github.com/pkg/errors",
+    "github.com/problame/go-netssh",
+    "github.com/problame/go-rwccmd",
+    "github.com/problame/go-streamrpc",
+    "github.com/prometheus/client_golang/prometheus",
+    "github.com/prometheus/client_golang/prometheus/promhttp",
+    "github.com/spf13/cobra",
+    "github.com/spf13/pflag",
+    "github.com/stretchr/testify/assert",
+    "github.com/stretchr/testify/require",
+    "github.com/zrepl/yaml-config",
+    "golang.org/x/sys/unix",
+  ]
+  solver-name = "gps-cdcl"
+  solver-version = 1

Gopkg.toml

@@ -0,0 +1,75 @@
+ignored = [ "github.com/inconshreveable/mousetrap" ]
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/ftrvxmtrx/fd"
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/jinzhu/copier"
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/kr/pretty"
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/mitchellh/go-homedir"
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/mitchellh/mapstructure"
+
+[[constraint]]
+  name = "github.com/pkg/errors"
+  version = "0.8.0"
+
+[[constraint]]
+  branch = "master"
+  name = "github.com/spf13/cobra"
+
+[[constraint]]
+  name = "github.com/spf13/viper"
+  version = "1.0.0"
+
+[[constraint]]
+  name = "github.com/stretchr/testify"
+  version = "1.1.4"
+
+[[constraint]]
+  branch = "v2"
+  name = "github.com/zrepl/yaml-config"
+
+[[constraint]]
+  name = "github.com/go-logfmt/logfmt"
+  version = "*"
+
+[[constraint]]
+  name = "github.com/problame/go-rwccmd"
+  branch = "master"
+
+[[constraint]]
+  name = "github.com/problame/go-netssh"
+  branch = "master"
+
+[[constraint]]
+  name = "github.com/prometheus/client_golang"
+  branch = "master"
+
+[[constraint]]
+  name = "github.com/golang/protobuf"
+  version = "1.2.0"
+
+[[constraint]]
+  name = "github.com/fatih/color"
+  version = "1.7.0"
+
+[[constraint]]
+  name = "github.com/problame/go-streamrpc"
+  version = "0.5.0"
+
+
+
+[[constraint]]
+  name = "github.com/gdamore/tcell"
+  version = "1.0.0"

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2017 Christian Schwarz
+Copyright (c) 2017-2018 Christian Schwarz
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

Makefile

@@ -0,0 +1,143 @@
+.PHONY: generate build test vet cover release docs docs-clean clean vendordeps
+.DEFAULT_GOAL := build
+
+ROOT := github.com/zrepl/zrepl
+SUBPKGS += client
+SUBPKGS += config
+SUBPKGS += daemon
+SUBPKGS += daemon/filters
+SUBPKGS += daemon/job
+SUBPKGS += daemon/logging
+SUBPKGS += daemon/nethelpers
+SUBPKGS += daemon/pruner
+SUBPKGS += daemon/snapper
+SUBPKGS += daemon/streamrpcconfig
+SUBPKGS += daemon/transport
+SUBPKGS += daemon/transport/connecter
+SUBPKGS += daemon/transport/serve
+SUBPKGS += endpoint
+SUBPKGS += logger
+SUBPKGS += pruning
+SUBPKGS += pruning/retentiongrid
+SUBPKGS += replication
+SUBPKGS += replication/fsrep
+SUBPKGS += replication/pdu
+SUBPKGS += replication/internal/diff
+SUBPKGS += tlsconf
+SUBPKGS += util
+SUBPKGS += util/socketpair
+SUBPKGS += util/watchdog
+SUBPKGS += util/envconst
+SUBPKGS += version
+SUBPKGS += zfs
+
+_TESTPKGS := $(ROOT) $(foreach p,$(SUBPKGS),$(ROOT)/$(p))
+
+ARTIFACTDIR := artifacts
+
+ifdef ZREPL_VERSION
+    _ZREPL_VERSION := $(ZREPL_VERSION)
+endif
+ifndef _ZREPL_VERSION
+    _ZREPL_VERSION := $(shell git describe --dirty 2>/dev/null || echo "ZREPL_BUILD_INVALID_VERSION" )
+    ifeq ($(_ZREPL_VERSION),ZREPL_BUILD_INVALID_VERSION) # can't use .SHELLSTATUS because Debian Stretch is still on gmake 4.1
+        $(error cannot infer variable ZREPL_VERSION using git and variable is not overriden by make invocation)
+    endif
+endif
+GO_LDFLAGS := "-X github.com/zrepl/zrepl/version.zreplVersion=$(_ZREPL_VERSION)"
+
+GO_BUILD := go build -ldflags $(GO_LDFLAGS)
+
+RELEASE_BINS := $(ARTIFACTDIR)/zrepl-freebsd-amd64
+RELEASE_BINS += $(ARTIFACTDIR)/zrepl-linux-amd64
+RELEASE_BINS += $(ARTIFACTDIR)/zrepl-darwin-amd64
+
+RELEASE_NOARCH := $(ARTIFACTDIR)/zrepl-noarch.tar
+THIS_PLATFORM_RELEASE_BIN := $(shell bash -c 'source <(go env) && echo "zrepl-$${GOOS}-$${GOARCH}"' )
+
+vendordeps:
+	dep ensure -v -vendor-only
+
+generate: #not part of the build, must do that manually
+	protoc -I=replication/pdu --go_out=replication/pdu replication/pdu/pdu.proto
+	@for pkg in $(_TESTPKGS); do\
+		go generate "$$pkg" || exit 1; \
+	done;
+
+build:
+	@echo "INFO: In case of missing dependencies, run 'make vendordeps'"
+	$(GO_BUILD) -o "$(ARTIFACTDIR)/zrepl"
+
+test:
+	@for pkg in $(_TESTPKGS); do \
+		echo "Testing $$pkg"; \
+		go test "$$pkg" || exit 1; \
+	done;
+
+vet:
+	@for pkg in $(_TESTPKGS); do \
+		echo "Vetting $$pkg"; \
+		go vet "$$pkg" || exit 1; \
+	done;
+
+cover: artifacts
+	@for pkg in $(_TESTPKGS); do \
+		profile="$(ARTIFACTDIR)/cover-$$(basename $$pkg).out"; \
+		go test -coverprofile "$$profile" $$pkg || exit 1; \
+		if [ -f "$$profile" ]; then \
+   			go tool cover -html="$$profile" -o "$${profile}.html" || exit 2; \
+		fi; \
+	done;
+
+$(ARTIFACTDIR):
+	mkdir -p "$@"
+
+$(ARTIFACTDIR)/docs: $(ARTIFACTDIR)
+	mkdir -p "$@"
+
+$(ARTIFACTDIR)/bash_completion: $(RELEASE_BINS)
+	artifacts/$(THIS_PLATFORM_RELEASE_BIN) bashcomp "$@"
+
+$(ARTIFACTDIR)/go_version.txt:
+	go version > $@
+
+docs: $(ARTIFACTDIR)/docs
+	make -C docs \
+		html \
+		BUILDDIR=../artifacts/docs \
+
+docs-clean:
+	make -C docs \
+		clean \
+		BUILDDIR=../artifacts/docs
+
+
+.PHONY: $(RELEASE_BINS)
+# TODO: two wildcards possible
+$(RELEASE_BINS): $(ARTIFACTDIR)/zrepl-%-amd64: generate $(ARTIFACTDIR) vet test
+	@echo "INFO: In case of missing dependencies, run 'make vendordeps'"
+	GOOS=$* GOARCH=amd64   $(GO_BUILD) -o "$(ARTIFACTDIR)/zrepl-$*-amd64"
+
+$(RELEASE_NOARCH): docs $(ARTIFACTDIR)/bash_completion $(ARTIFACTDIR)/go_version.txt
+	tar --mtime='1970-01-01' --sort=name \
+		--transform 's/$(ARTIFACTDIR)/zrepl-$(_ZREPL_VERSION)-noarch/' \
+		-acf $@ \
+		$(ARTIFACTDIR)/docs/html \
+		$(ARTIFACTDIR)/bash_completion \
+		$(ARTIFACTDIR)/go_version.txt
+
+release: $(RELEASE_BINS) $(RELEASE_NOARCH)
+	rm -rf "$(ARTIFACTDIR)/release"
+	mkdir -p "$(ARTIFACTDIR)/release"
+	cp $^ "$(ARTIFACTDIR)/release"
+	cd "$(ARTIFACTDIR)/release" && sha512sum $$(ls | sort) > sha512sum.txt
+	@# note that we use ZREPL_VERSION and not _ZREPL_VERSION because we want to detect the override
+	@if git describe --dirty 2>/dev/null | grep dirty >/dev/null; then \
+		if [ "$(ZREPL_VERSION)" == "" ]; then \
+			echo "[WARN] git checkout is dirty and make variable ZREPL_VERSION was not used to override"; \
+			exit 1; \
+		fi; \
+	fi;
+
+clean: docs-clean
+	rm -rf "$(ARTIFACTDIR)"

README.md

@@ -1,2 +1,94 @@
 # zrepl
-ZFS dataset replication. Not a shell script.
+zrepl is a one-stop ZFS backup & replication solution.
+
+## User Documentation
+
+**User Documentation** can be found at [zrepl.github.io](https://zrepl.github.io).
+
+## Bug Reports
+
+1. If the issue is reproducible, enable debug logging, reproduce and capture the log.
+2. Open an issue on GitHub, with logs pasted as GitHub gists / inline.
+
+## Feature Requests
+
+1. Does you feature request require default values / some kind of configuration?
+   If so, think of an expressive configuration example.
+2. Think of at least one use case that generalizes from your concrete application.
+3. Open an issue on GitHub with example conf & use case attached.
+
+The above does not apply if you already implemented everything.
+Check out the *Coding Workflow* section below for details.
+
+## Package Maintainer Information
+
+* Follow the steps in `docs/installation.rst -> Compiling from Source` and read the Makefile / shell scripts used in this process.
+* Make sure your distro is compatible with the paths in `docs/installation.rst`.
+* Ship a default config that adheres to your distro's `hier` and logging system.
+* Ship a service manager file and _please_ try to upstream it to this repository.
+* Use `make release ZREPL_VERSION='mydistro-1.2.3_1'`
+    * Your distro's name and any versioning supplemental to zrepl's (e.g. package revision) should be in this string
+* Make sure you are informed about new zrepl versions, e.g. by subscribing to GitHub's release RSS feed.
+
+## Developer Documentation
+
+zrepl is written in [Go](https://golang.org) and uses [`dep`](https://github.com/golang/dep) to manage dependencies.
+The documentation is written in [ReStructured Text](http://docutils.sourceforge.net/rst.html) using the [Sphinx](https://www.sphinx-doc.org) framework.
+
+To get started, run `./lazy.sh devsetup` to easily install build dependencies and read `docs/installation.rst -> Compiling from Source`.
+
+### Overall Architecture
+
+The application architecture is documented as part of the user docs in the *Implementation* section (`docs/content/impl`).
+Make sure to develop an understanding how zrepl is typically used by studying the user docs first.
+
+### Project Structure
+
+```
+├── cmd
+│   ├── endpoint            # implementations of endpoints for package replication
+│   ├── sampleconf          # example configuration
+├── docs                    # sphinx-based documentation
+│   ├── **/*.rst            # documentation in reStructuredText
+│   ├── sphinxconf
+│   │   └── conf.py         # sphinx config (see commit 445a280 why its not in docs/)
+│   ├── requirements.txt    # pip3 requirements to build documentation
+│   ├── publish.sh          # shell script for automated rendering & deploy to zrepl.github.io repo
+│   ├── public_git          # checkout of zrepl.github.io managed by above shell script
+├── logger                  # logger package used by zrepl
+├── replication             # replication functionality
+├── rpc                     # rpc protocol implementation
+├── util
+└── zfs                     # ZFS wrappers, filesystemm diffing
+```
+
+### Coding Workflow
+
+* Open an issue when starting to hack on a new feature
+* Commits should reference the issue they are related to
+* Docs improvements not documenting new features do not require an issue.
+
+### Breaking Changes
+
+Backward-incompatible changes must be documented in the git commit message and are listed in `docs/changelog.rst`.
+
+* Config-breaking changes must contain a line `BREAK CONFIG` in the commit message
+* Other breaking changes must contain a line `BREAK` in the commit message
+
+### Glossary & Naming Inconsistencies
+
+In ZFS, *dataset* refers to the objects *filesystem*, *ZVOL* and *snapshot*. <br />
+However, we need a word for *filesystem* & *ZVOL* but not a snapshot, bookmark, etc.
+
+Toward the user, the following terminology is used:
+
+* **filesystem**: a ZFS filesystem or a ZVOL
+* **filesystem version**: a ZFS snapshot or a bookmark
+
+Sadly, the zrepl implementation is inconsistent in its use of these words:
+variables and types are often named *dataset* when they in fact refer to a *filesystem*.
+
+There will not be a big refactoring (an attempt was made, but it's destroying too much history without much gain).
+
+However, new contributions & patches should fix naming without further notice in the commit message.
+

build.Dockerfile

@@ -0,0 +1,22 @@
+FROM golang:latest
+
+RUN apt-get update && apt-get install -y \
+    python3-pip \
+    unzip
+
+RUN wget https://github.com/protocolbuffers/protobuf/releases/download/v3.6.1/protoc-3.6.1-linux-x86_64.zip
+RUN echo "6003de742ea3fcf703cfec1cd4a3380fd143081a2eb0e559065563496af27807  protoc-3.6.1-linux-x86_64.zip" | sha256sum -c
+RUN unzip -d /usr protoc-3.6.1-linux-x86_64.zip
+
+ADD lazy.sh /tmp/lazy.sh
+ADD docs/requirements.txt /tmp/requirements.txt
+ENV ZREPL_LAZY_DOCS_REQPATH=/tmp/requirements.txt
+RUN /tmp/lazy.sh devsetup
+
+# prepare volume mount of git checkout to /zrepl
+RUN mkdir -p /go/src/github.com/zrepl/zrepl
+RUN chmod -R 0777 /go
+RUN mkdir -p /.cache && chmod -R 0777 /.cache
+WORKDIR /go/src/github.com/zrepl/zrepl
+
+

cli/cli.go

@@ -0,0 +1,116 @@
+package cli
+
+import (
+	"fmt"
+	"github.com/spf13/cobra"
+	"github.com/spf13/pflag"
+	"github.com/zrepl/zrepl/config"
+	"os"
+)
+
+var rootArgs struct {
+	configPath string
+}
+
+var rootCmd = &cobra.Command{
+	Use:   "zrepl",
+	Short: "One-stop ZFS replication solution",
+}
+
+var bashcompCmd = &cobra.Command{
+	Use:   "bashcomp path/to/out/file",
+	Short: "generate bash completions",
+	Run: func(cmd *cobra.Command, args []string) {
+		if len(args) != 1 {
+			fmt.Fprintf(os.Stderr, "specify exactly one positional agument\n")
+			cmd.Usage()
+			os.Exit(1)
+		}
+		if err := rootCmd.GenBashCompletionFile(args[0]); err != nil {
+			fmt.Fprintf(os.Stderr, "error generating bash completion: %s", err)
+			os.Exit(1)
+		}
+	},
+	Hidden: true,
+}
+
+func init() {
+	rootCmd.PersistentFlags().StringVar(&rootArgs.configPath, "config", "", "config file path")
+	rootCmd.AddCommand(bashcompCmd)
+}
+
+type Subcommand struct {
+	Use             string
+	Short           string
+	NoRequireConfig bool
+	Run             func(subcommand *Subcommand, args []string) error
+	SetupFlags      func(f *pflag.FlagSet)
+	SetupSubcommands  func() []*Subcommand
+
+	config *config.Config
+	configErr error
+}
+
+func (s *Subcommand) ConfigParsingError() error {
+	return s.configErr
+}
+
+func (s *Subcommand) Config() *config.Config {
+	if !s.NoRequireConfig && s.config == nil {
+		panic("command that requires config is running and has no config set")
+	}
+	return s.config
+}
+
+func (s *Subcommand) run(cmd *cobra.Command, args []string) {
+	s.tryParseConfig()
+	err := s.Run(s, args)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "%s\n", err)
+		os.Exit(1)
+	}
+}
+
+func (s *Subcommand) tryParseConfig() {
+	config, err := config.ParseConfig(rootArgs.configPath)
+	s.configErr = err
+	if err != nil {
+		if s.NoRequireConfig {
+			// doesn't matter
+			return
+		} else {
+			fmt.Fprintf(os.Stderr, "could not parse config: %s\n", err)
+			os.Exit(1)
+		}
+	}
+	s.config = config
+}
+
+func AddSubcommand(s *Subcommand) {
+	addSubcommandToCobraCmd(rootCmd, s)
+}
+
+func addSubcommandToCobraCmd(c *cobra.Command, s *Subcommand) {
+	cmd := cobra.Command{
+		Use: s.Use,
+		Short: s.Short,
+	}
+	if s.SetupSubcommands == nil {
+		cmd.Run = s.run
+	} else {
+		for _, sub := range s.SetupSubcommands() {
+			addSubcommandToCobraCmd(&cmd, sub)
+		}
+	}
+	if s.SetupFlags != nil {
+		s.SetupFlags(cmd.Flags())
+	}
+	c.AddCommand(&cmd)
+}
+
+
+func Run() {
+	if err := rootCmd.Execute(); err != nil {
+		os.Exit(1)
+	}
+}

client/configcheck.go

@@ -0,0 +1,112 @@
+package client
+
+import (
+	"encoding/json"
+	"fmt"
+	"github.com/kr/pretty"
+	"github.com/pkg/errors"
+	"github.com/spf13/pflag"
+	"github.com/zrepl/yaml-config"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/job"
+	"github.com/zrepl/zrepl/daemon/logging"
+	"github.com/zrepl/zrepl/logger"
+	"os"
+)
+
+var configcheckArgs struct {
+	format string
+	what string
+}
+
+var ConfigcheckCmd = &cli.Subcommand{
+	Use: "configcheck",
+	Short: "check if config can be parsed without errors",
+	SetupFlags: func(f *pflag.FlagSet) {
+		f.StringVar(&configcheckArgs.format, "format", "", "dump parsed config object [pretty|yaml|json]")
+		f.StringVar(&configcheckArgs.what, "what", "all", "what to print [all|config|jobs|logging]")
+	},
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		formatMap := map[string]func(interface{}) {
+			"": func(i interface{}) {},
+			"pretty": func(i interface{}) { pretty.Println(i) },
+			"json": func(i interface{}) {
+				json.NewEncoder(os.Stdout).Encode(subcommand.Config())
+			},
+			"yaml": func(i interface{}) {
+				yaml.NewEncoder(os.Stdout).Encode(subcommand.Config())
+			},
+		}
+
+		formatter, ok := formatMap[configcheckArgs.format]
+		if !ok {
+			return fmt.Errorf("unsupported --format %q", configcheckArgs.format)
+		}
+
+		var hadErr bool
+		// further: try to build jobs
+		confJobs, err := job.JobsFromConfig(subcommand.Config())
+		if err != nil {
+			err := errors.Wrap(err, "cannot build jobs from config")
+			if configcheckArgs.what == "jobs" {
+				return err
+			} else {
+				fmt.Fprintf(os.Stderr, "%s\n", err)
+				confJobs = nil
+				hadErr = true
+			}
+		}
+
+		// further: try to build logging outlets
+		outlets, err := logging.OutletsFromConfig(*subcommand.Config().Global.Logging)
+		if err != nil {
+			err := errors.Wrap(err, "cannot build logging from config")
+			if configcheckArgs.what == "logging" {
+				return err
+			} else {
+				fmt.Fprintf(os.Stderr, "%s\n", err)
+				outlets = nil
+				hadErr = true
+			}
+		}
+
+
+		whatMap := map[string]func() {
+			"all": func() {
+				o := struct {
+					config *config.Config
+					jobs []job.Job
+					logging *logger.Outlets
+				}{
+					subcommand.Config(),
+					confJobs,
+					outlets,
+				}
+				formatter(o)
+			},
+			"config": func() {
+				formatter(subcommand.Config())
+			},
+			"jobs": func() {
+				formatter(confJobs)
+			},
+			"logging": func() {
+				formatter(outlets)
+			},
+		}
+
+		wf, ok := whatMap[configcheckArgs.what]
+		if !ok {
+			return fmt.Errorf("unsupported --format %q", configcheckArgs.what)
+		}
+		wf()
+
+		if hadErr {
+			return fmt.Errorf("config parsing failed")
+		} else {
+			return nil
+		}
+	},
+}
+

client/jsonclient.go

@@ -0,0 +1,48 @@
+package client
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"github.com/pkg/errors"
+	"io"
+	"net"
+	"net/http"
+)
+
+func controlHttpClient(sockpath string) (client http.Client, err error) {
+	return http.Client{
+		Transport: &http.Transport{
+			DialContext: func(_ context.Context, _, _ string) (net.Conn, error) {
+				return net.Dial("unix", sockpath)
+			},
+		},
+	}, nil
+}
+
+func jsonRequestResponse(c http.Client, endpoint string, req interface{}, res interface{}) error {
+	var buf bytes.Buffer
+	encodeErr := json.NewEncoder(&buf).Encode(req)
+	if encodeErr != nil {
+		return encodeErr
+	}
+
+	resp, err := c.Post("http://unix"+endpoint, "application/json", &buf)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		var msg bytes.Buffer
+		io.CopyN(&msg, resp.Body, 4096)
+		return errors.Errorf("%s", msg.String())
+	}
+
+	decodeError := json.NewDecoder(resp.Body).Decode(&res)
+	if decodeError != nil {
+		return decodeError
+	}
+
+	return nil
+}

client/pprof.go

@@ -0,0 +1,66 @@
+package client
+
+import (
+	"errors"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon"
+	"log"
+	"os"
+)
+
+var pprofArgs struct {
+	daemon.PprofServerControlMsg
+}
+
+var PprofCmd = &cli.Subcommand{
+	Use:   "pprof off | [on TCP_LISTEN_ADDRESS]",
+	Short: "start a http server exposing go-tool-compatible profiling endpoints at TCP_LISTEN_ADDRESS",
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		if len(args) < 1 {
+			goto enargs
+		}
+		switch args[0] {
+		case "on":
+			pprofArgs.Run = true
+			if len(args) != 2 {
+				return errors.New("must specify TCP_LISTEN_ADDRESS as second positional argument")
+			}
+			pprofArgs.HttpListenAddress = args[1]
+		case "off":
+			if len(args) != 1 {
+				goto enargs
+			}
+			pprofArgs.Run = false
+		}
+
+		RunPProf(subcommand.Config())
+		return nil
+	enargs:
+		return errors.New("invalid number of positional arguments")
+
+	},
+}
+
+func RunPProf(conf *config.Config) {
+	log := log.New(os.Stderr, "", 0)
+
+	die := func() {
+		log.Printf("exiting after error")
+		os.Exit(1)
+	}
+
+	log.Printf("connecting to zrepl daemon")
+
+	httpc, err := controlHttpClient(conf.Global.Control.SockPath)
+	if err != nil {
+		log.Printf("error creating http client: %s", err)
+		die()
+	}
+	err = jsonRequestResponse(httpc, daemon.ControlJobEndpointPProf, pprofArgs.PprofServerControlMsg, struct{}{})
+	if err != nil {
+		log.Printf("error sending control message: %s", err)
+		die()
+	}
+	log.Printf("finished")
+}

client/signal.go

@@ -0,0 +1,39 @@
+package client
+
+import (
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon"
+)
+
+var SignalCmd = &cli.Subcommand{
+	Use:   "signal [wakeup|reset] JOB",
+	Short: "wake up a job from wait state or abort its current invocation",
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		return runSignalCmd(subcommand.Config(), args)
+	},
+}
+
+func runSignalCmd(config *config.Config, args []string) error {
+	if len(args) != 2 {
+		return errors.Errorf("Expected 2 arguments: [wakeup|reset] JOB")
+	}
+
+	httpc, err := controlHttpClient(config.Global.Control.SockPath)
+	if err != nil {
+		return err
+	}
+
+	err = jsonRequestResponse(httpc, daemon.ControlJobEndpointSignal,
+		struct {
+			Name string
+			Op string
+		}{
+			Name: args[1],
+			Op: args[0],
+		},
+		struct{}{},
+	)
+	return err
+}

client/status.go

@@ -0,0 +1,633 @@
+package client
+
+import (
+	"fmt"
+	"github.com/gdamore/tcell/termbox"
+	"github.com/pkg/errors"
+	"github.com/spf13/pflag"
+	"github.com/zrepl/yaml-config"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/daemon"
+	"github.com/zrepl/zrepl/daemon/job"
+	"github.com/zrepl/zrepl/daemon/pruner"
+	"github.com/zrepl/zrepl/replication"
+	"github.com/zrepl/zrepl/replication/fsrep"
+	"io"
+	"math"
+	"net/http"
+	"os"
+	"sort"
+	"strings"
+	"sync"
+	"time"
+)
+
+type bytesProgressHistory struct {
+	changeCounter     int
+	lastChangeAt      time.Time
+	last              int64
+	bpsIncreaseExpAvg float64
+}
+
+func (p *bytesProgressHistory) Update(currentVal int64) (bytesPerSecondAvg int64, changeCount int) {
+	if currentVal < p.last {
+		*p = bytesProgressHistory{
+			last: currentVal,
+			lastChangeAt: time.Now(),
+		}
+	}
+	defer func() {
+		p.last = currentVal
+	}()
+	if time.Now().Sub(p.lastChangeAt) > 3 *time.Second { // FIXME depends on refresh frequency
+		p.changeCounter = 0
+		p.bpsIncreaseExpAvg = 0
+	}
+	if currentVal != p.last {
+		p.changeCounter++
+		p.lastChangeAt = time.Now()
+	}
+
+	byteIncrease := float64(currentVal - p.last)
+	if byteIncrease < 0 {
+		byteIncrease = 0
+	}
+	const factor = 0.1
+	p.bpsIncreaseExpAvg = (1-factor) * p.bpsIncreaseExpAvg + factor *byteIncrease
+	return int64(p.bpsIncreaseExpAvg), p.changeCounter
+}
+
+type tui struct {
+	x, y   int
+	indent int
+
+	lock   sync.Mutex //For report and error
+	report map[string]job.Status
+	err    error
+
+	replicationProgress map[string]*bytesProgressHistory // by job name
+}
+
+func newTui() tui {
+	return tui{
+		replicationProgress: make(map[string]*bytesProgressHistory, 0),
+	}
+}
+
+func (t *tui) moveCursor(x, y int) {
+	t.x += x
+	t.y += y
+}
+
+const INDENT_MULTIPLIER = 4
+
+func (t *tui) moveLine(dl int, col int) {
+	t.y += dl
+	t.x = t.indent*INDENT_MULTIPLIER + col
+}
+
+func (t *tui) write(text string) {
+	for _, c := range text {
+		if c == '\n' {
+			t.newline()
+			continue
+		}
+		termbox.SetCell(t.x, t.y, c, termbox.ColorDefault, termbox.ColorDefault)
+		t.x += 1
+	}
+}
+
+func (t *tui) printf(text string, a ...interface{}) {
+	t.write(fmt.Sprintf(text, a...))
+}
+
+func wrap(s string, width int) string {
+	var b strings.Builder
+	for len(s) > 0 {
+		rem := width
+		if rem > len(s) {
+			rem = len(s)
+		}
+		if idx := strings.IndexAny(s, "\n\r"); idx != -1 && idx < rem {
+			rem = idx+1
+		}
+		untilNewline := strings.TrimSpace(s[:rem])
+		s = s[rem:]
+		if len(untilNewline) == 0 {
+			continue
+		}
+		b.WriteString(untilNewline)
+		b.WriteString("\n")
+	}
+	return strings.TrimSpace(b.String())
+}
+
+func (t *tui) printfDrawIndentedAndWrappedIfMultiline(format string, a ...interface{}) {
+	whole := fmt.Sprintf(format, a...)
+	width, _ := termbox.Size()
+	if !strings.ContainsAny(whole, "\n\r") && t.x + len(whole) <= width {
+		t.printf(format, a...)
+	} else {
+		t.addIndent(1)
+		t.newline()
+		t.write(wrap(whole, width - INDENT_MULTIPLIER*t.indent))
+		t.addIndent(-1)
+	}
+}
+
+func (t *tui) newline() {
+	t.moveLine(1, 0)
+}
+
+func (t *tui) setIndent(indent int) {
+	t.indent = indent
+	t.moveLine(0, 0)
+}
+
+func (t *tui) addIndent(indent int) {
+	t.indent += indent
+	t.moveLine(0, 0)
+}
+
+
+var statusFlags struct {
+	Raw bool
+}
+
+var StatusCmd = &cli.Subcommand{
+	Use:   "status",
+	Short: "show job activity or dump as JSON for monitoring",
+	SetupFlags: func(f *pflag.FlagSet) {
+		f.BoolVar(&statusFlags.Raw, "raw", false, "dump raw status description from zrepl daemon")
+	},
+	Run: runStatus,
+}
+
+func runStatus(s *cli.Subcommand, args []string) error {
+	httpc, err := controlHttpClient(s.Config().Global.Control.SockPath)
+	if err != nil {
+		return err
+	}
+
+	if statusFlags.Raw {
+		resp, err := httpc.Get("http://unix"+daemon.ControlJobEndpointStatus)
+		if err != nil {
+			return err
+		}
+		defer resp.Body.Close()
+		if resp.StatusCode != http.StatusOK {
+			fmt.Fprintf(os.Stderr, "Received error response:\n")
+			io.CopyN(os.Stderr, resp.Body, 4096)
+			return errors.Errorf("exit")
+		}
+		if _, err := io.Copy(os.Stdout, resp.Body); err != nil {
+			return err
+		}
+		return nil
+	}
+
+	t := newTui()
+	t.lock.Lock()
+	t.err = errors.New("Got no report yet")
+	t.lock.Unlock()
+
+	err = termbox.Init()
+	if err != nil {
+		return err
+	}
+	defer termbox.Close()
+
+	update := func() {
+		m := make(map[string]job.Status)
+
+		err2 := jsonRequestResponse(httpc, daemon.ControlJobEndpointStatus,
+			struct{}{},
+			&m,
+		)
+
+		t.lock.Lock()
+		t.err = err2
+		t.report = m
+		t.lock.Unlock()
+		t.draw()
+	}
+	update()
+
+	ticker := time.NewTicker(500 * time.Millisecond)
+	defer ticker.Stop()
+	go func() {
+		for _ = range ticker.C {
+			update()
+		}
+	}()
+
+	termbox.HideCursor()
+	termbox.Clear(termbox.ColorDefault, termbox.ColorDefault)
+
+loop:
+	for {
+		switch ev := termbox.PollEvent(); ev.Type {
+		case termbox.EventKey:
+			switch ev.Key {
+			case termbox.KeyEsc:
+				break loop
+			case termbox.KeyCtrlC:
+				break loop
+			}
+		case termbox.EventResize:
+			t.draw()
+		}
+	}
+
+	return nil
+
+}
+
+func (t *tui) getReplicationProgresHistory(jobName string) *bytesProgressHistory {
+	p, ok := t.replicationProgress[jobName]
+	if !ok {
+		p = &bytesProgressHistory{}
+		t.replicationProgress[jobName] = p
+	}
+	return p
+}
+
+func (t *tui) draw() {
+	t.lock.Lock()
+	defer t.lock.Unlock()
+
+	termbox.Clear(termbox.ColorDefault, termbox.ColorDefault)
+	t.x = 0
+	t.y = 0
+	t.indent = 0
+
+	if t.err != nil {
+		t.write(t.err.Error())
+	} else {
+		//Iterate over map in alphabetical order
+		keys := make([]string, len(t.report))
+		i := 0
+		for k, _ := range t.report {
+			keys[i] = k
+			i++
+		}
+		sort.Strings(keys)
+
+		for _, k := range keys {
+			v := t.report[k]
+			if len(k) == 0 || daemon.IsInternalJobName(k) { //Internal job
+				continue
+			}
+			t.setIndent(0)
+
+			t.printf("Job: %s", k)
+			t.setIndent(1)
+			t.newline()
+			t.printf("Type: %s", v.Type)
+			t.setIndent(1)
+			t.newline()
+
+			if v.Type != job.TypePush && v.Type != job.TypePull {
+				t.printf("No status representation for job type '%s', dumping as YAML", v.Type)
+				t.newline()
+				asYaml, err := yaml.Marshal(v.JobSpecific)
+				if err != nil {
+					t.printf("Error marshaling status to YAML: %s", err)
+					t.newline()
+					continue
+				}
+				t.write(string(asYaml))
+				t.newline()
+				continue
+			}
+
+			pushStatus, ok := v.JobSpecific.(*job.ActiveSideStatus)
+			if !ok || pushStatus == nil {
+				t.printf("ActiveSideStatus is null")
+				t.newline()
+				continue
+			}
+
+			t.printf("Replication:")
+			t.newline()
+			t.addIndent(1)
+			t.renderReplicationReport(pushStatus.Replication, t.getReplicationProgresHistory(k))
+			t.addIndent(-1)
+
+			t.printf("Pruning Sender:")
+			t.newline()
+			t.addIndent(1)
+			t.renderPrunerReport(pushStatus.PruningSender)
+			t.addIndent(-1)
+
+			t.printf("Pruning Receiver:")
+			t.newline()
+			t.addIndent(1)
+			t.renderPrunerReport(pushStatus.PruningReceiver)
+			t.addIndent(-1)
+
+		}
+	}
+	termbox.Flush()
+}
+
+func (t *tui) renderReplicationReport(rep *replication.Report, history *bytesProgressHistory) {
+	if rep == nil {
+		t.printf("...\n")
+		return
+	}
+
+	all := make([]*fsrep.Report, 0, len(rep.Completed)+len(rep.Pending) + 1)
+	all = append(all, rep.Completed...)
+	all = append(all, rep.Pending...)
+	if rep.Active != nil {
+		all = append(all, rep.Active)
+	}
+	sort.Slice(all, func(i, j int) bool {
+		return all[i].Filesystem < all[j].Filesystem
+	})
+
+	state, err := replication.StateString(rep.Status)
+	if err != nil {
+		t.printf("Status: %q (parse error: %q)\n", rep.Status, err)
+		return
+	}
+
+	t.printf("Status: %s", state)
+	t.newline()
+	if rep.Problem != "" {
+		t.printf("Problem: ")
+		t.printfDrawIndentedAndWrappedIfMultiline("%s", rep.Problem)
+		t.newline()
+	}
+	if rep.SleepUntil.After(time.Now()) && !state.IsTerminal() {
+		t.printf("Sleeping until %s (%s left)\n", rep.SleepUntil, rep.SleepUntil.Sub(time.Now()))
+	}
+
+	if state != replication.Planning && state != replication.PlanningError {
+		// Progress: [---------------]
+		sumUpFSRep := func(rep *fsrep.Report) (transferred, total int64) {
+			for _, s := range rep.Pending {
+				transferred += s.Bytes
+				total  += s.ExpectedBytes
+			}
+			for _, s := range rep.Completed {
+				transferred += s.Bytes
+				total += s.ExpectedBytes
+			}
+			return
+		}
+		var transferred, total int64
+		for _, fs := range all {
+			fstx, fstotal := sumUpFSRep(fs)
+			transferred += fstx
+			total += fstotal
+		}
+		rate, changeCount := history.Update(transferred)
+		t.write("Progress: ")
+		t.drawBar(50, transferred, total, changeCount)
+		t.write(fmt.Sprintf(" %s / %s @ %s/s", ByteCountBinary(transferred), ByteCountBinary(total), ByteCountBinary(rate)))
+		t.newline()
+	}
+
+	var maxFSLen int
+	for _, fs := range all {
+		if len(fs.Filesystem) > maxFSLen {
+			maxFSLen = len(fs.Filesystem)
+		}
+	}
+	for _, fs := range all {
+		t.printFilesystemStatus(fs, fs == rep.Active, maxFSLen)
+	}
+}
+
+func (t *tui) renderPrunerReport(r *pruner.Report) {
+	if r == nil {
+		t.printf("...\n")
+		return
+	}
+
+	state, err := pruner.StateString(r.State)
+	if err != nil {
+		t.printf("Status: %q (parse error: %q)\n", r.State, err)
+		return
+	}
+
+	t.printf("Status: %s", state)
+	t.newline()
+
+	if r.Error != "" {
+		t.printf("Error: %s\n", r.Error)
+	}
+	if r.SleepUntil.After(time.Now()) {
+		t.printf("Sleeping until %s (%s left)\n", r.SleepUntil, r.SleepUntil.Sub(time.Now()))
+	}
+
+	type commonFS struct {
+		*pruner.FSReport
+		completed bool
+	}
+	all := make([]commonFS, 0, len(r.Pending) + len(r.Completed))
+	for i := range r.Pending {
+		all = append(all, commonFS{&r.Pending[i], false})
+	}
+	for i := range r.Completed {
+		all = append(all, commonFS{&r.Completed[i], true})
+	}
+
+	switch state {
+	case pruner.Plan: fallthrough
+	case pruner.PlanWait: fallthrough
+	case pruner.ErrPerm:
+		return
+	}
+
+	if len(all) == 0 {
+		t.printf("nothing to do\n")
+		return
+	}
+
+	var totalDestroyCount, completedDestroyCount int
+	var maxFSname int
+	for _, fs := range all {
+		totalDestroyCount += len(fs.DestroyList)
+		if fs.completed {
+			completedDestroyCount += len(fs.DestroyList)
+		}
+		if maxFSname < len(fs.Filesystem) {
+			maxFSname = len(fs.Filesystem)
+		}
+	}
+
+	// global progress bar
+	progress := int(math.Round(80 * float64(completedDestroyCount) / float64(totalDestroyCount)))
+	t.write("Progress: ")
+	t.write("[")
+	t.write(times("=", progress))
+	t.write(">")
+	t.write(times("-", 80 - progress))
+	t.write("]")
+	t.printf(" %d/%d snapshots", completedDestroyCount, totalDestroyCount)
+	t.newline()
+
+	sort.SliceStable(all, func(i, j int) bool {
+		return strings.Compare(all[i].Filesystem, all[j].Filesystem) == -1
+	})
+
+	// Draw a table-like representation of 'all'
+	for _, fs := range all {
+		t.write(rightPad(fs.Filesystem, maxFSname, " "))
+		t.write(" ")
+		if fs.LastError != "" {
+			t.printf("ERROR (%d): %s\n", fs.ErrorCount, fs.LastError) // whitespace is padding
+			continue
+		}
+
+		pruneRuleActionStr := fmt.Sprintf("(destroy %d of %d snapshots)",
+			len(fs.DestroyList), len(fs.SnapshotList))
+
+		if fs.completed {
+			t.printf( "Completed  %s\n", pruneRuleActionStr)
+			continue
+		}
+
+		t.write("Pending    ") // whitespace is padding 10
+		if len(fs.DestroyList) == 1 {
+			t.write(fs.DestroyList[0].Name)
+		} else {
+			t.write(pruneRuleActionStr)
+		}
+		t.newline()
+	}
+
+}
+
+const snapshotIndent = 1
+func calculateMaxFSLength(all []*fsrep.Report) (maxFS, maxStatus int) {
+	for _, e := range all {
+		if len(e.Filesystem) > maxFS {
+			maxFS = len(e.Filesystem)
+		}
+		all2 := make([]*fsrep.StepReport, 0, len(e.Pending) + len(e.Completed))
+		all2 = append(all2, e.Pending...)
+		all2 = append(all2, e.Completed...)
+		for _, e2 := range all2 {
+			elen := len(e2.Problem) + len(e2.From) + len(e2.To) + 60 // random spacing, units, labels, etc
+			if elen > maxStatus {
+				maxStatus = elen
+			}
+		}
+	}
+	return
+}
+
+func times(str string, n int) (out string) {
+	for i := 0; i < n; i++ {
+		out += str
+	}
+	return
+}
+
+func rightPad(str string, length int, pad string) string {
+	if len(str) > length {
+		return str[:length]
+	}
+	return str + times(pad, length-len(str))
+}
+
+
+func leftPad(str string, length int, pad string) string {
+	if len(str) > length {
+		return str[len(str)-length:]
+	}
+	return times(pad, length-len(str)) + str
+}
+
+var arrowPositions = `>\|/`
+
+// changeCount = 0 indicates stall / no progresss
+func (t *tui) drawBar(length int, bytes, totalBytes int64, changeCount int) {
+	var completedLength int
+	if totalBytes > 0 {
+		completedLength = int(int64(length) * bytes / totalBytes)
+		if completedLength > length {
+			completedLength = length
+		}
+	} else if totalBytes == bytes {
+		completedLength = length
+	}
+
+	t.write("[")
+	t.write(times("=", completedLength))
+	t.write( string(arrowPositions[changeCount%len(arrowPositions)]))
+	t.write(times("-", length-completedLength))
+	t.write("]")
+}
+
+func StringStepState(s fsrep.StepState) string {
+	switch s {
+	case fsrep.StepReplicationReady: return "Ready"
+	case fsrep.StepMarkReplicatedReady: return "MarkReady"
+	case fsrep.StepCompleted: return "Completed"
+	default:
+		return fmt.Sprintf("UNKNOWN %d", s)
+	}
+}
+
+func (t *tui) printFilesystemStatus(rep *fsrep.Report, active bool, maxFS int) {
+
+	bytes := int64(0)
+	totalBytes := int64(0)
+	for _, s := range rep.Pending {
+		bytes += s.Bytes
+		totalBytes += s.ExpectedBytes
+	}
+	for _, s := range rep.Completed {
+		bytes += s.Bytes
+		totalBytes += s.ExpectedBytes
+	}
+
+
+	status := fmt.Sprintf("%s (step %d/%d, %s/%s)",
+		rep.Status,
+		len(rep.Completed), len(rep.Pending) + len(rep.Completed),
+		ByteCountBinary(bytes), ByteCountBinary(totalBytes),
+
+	)
+
+	activeIndicator := " "
+	if active {
+		activeIndicator = "*"
+	}
+	t.printf("%s %s %s ",
+		activeIndicator,
+		rightPad(rep.Filesystem, maxFS, " "),
+		status)
+
+	next := ""
+	if rep.Problem != "" {
+		next = rep.Problem
+	} else if len(rep.Pending) > 0 {
+		if rep.Pending[0].From != "" {
+			next = fmt.Sprintf("next: %s => %s", rep.Pending[0].From, rep.Pending[0].To)
+		} else {
+			next = fmt.Sprintf("next: %s (full)", rep.Pending[0].To)
+		}
+	}
+	t.printfDrawIndentedAndWrappedIfMultiline("%s", next)
+
+	t.newline()
+}
+
+func ByteCountBinary(b int64) string {
+	const unit = 1024
+	if b < unit {
+		return fmt.Sprintf("%d B", b)
+	}
+	div, exp := int64(unit), 0
+	for n := b / unit; n >= unit; n /= unit {
+		div *= unit
+		exp++
+	}
+	return fmt.Sprintf("%.1f %ciB", float64(b)/float64(div), "KMGTPE"[exp])
+}

client/stdinserver.go

@@ -0,0 +1,49 @@
+package client
+
+import (
+	"github.com/zrepl/zrepl/cli"
+	"os"
+
+	"context"
+	"errors"
+	"github.com/problame/go-netssh"
+	"github.com/zrepl/zrepl/config"
+	"log"
+	"path"
+)
+
+var StdinserverCmd = &cli.Subcommand{
+	Use:   "stdinserver CLIENT_IDENTITY",
+	Short: "stdinserver transport mode (started from authorized_keys file as forced command)",
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		return runStdinserver(subcommand.Config(), args)
+	},
+}
+
+func runStdinserver(config *config.Config, args []string) error {
+
+	// NOTE: the netssh proxying protocol requires exiting with non-zero status if anything goes wrong
+	defer os.Exit(1)
+
+	log := log.New(os.Stderr, "", log.LUTC|log.Ldate|log.Ltime)
+
+	if len(args) != 1 || args[0] == "" {
+		err := errors.New("must specify client_identity as positional argument")
+		return err
+	}
+
+	identity := args[0]
+	unixaddr := path.Join(config.Global.Serve.StdinServer.SockDir, identity)
+
+	log.Printf("proxying client identity '%s' to zrepl daemon '%s'", identity, unixaddr)
+
+	ctx := netssh.ContextWithLog(context.TODO(), log)
+
+	err := netssh.Proxy(ctx, unixaddr)
+	if err == nil {
+		log.Print("proxying finished successfully, exiting with status 0")
+		os.Exit(0)
+	}
+	log.Printf("error proxying: %s", err)
+	return nil
+}

client/testcmd.go

@@ -0,0 +1,108 @@
+package client
+
+import (
+	"fmt"
+	"github.com/spf13/pflag"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/filters"
+	"github.com/zrepl/zrepl/zfs"
+)
+
+var TestCmd = &cli.Subcommand {
+	Use: "test",
+	SetupSubcommands: func() []*cli.Subcommand {
+		return []*cli.Subcommand{testFilter}
+	},
+}
+
+var testFilterArgs struct {
+	job string
+	all bool
+	input string
+}
+
+var testFilter = &cli.Subcommand{
+	Use: "filesystems --job JOB [--all | --input INPUT]",
+	Short: "test filesystems filter specified in push or source job",
+	SetupFlags: func(f *pflag.FlagSet) {
+		f.StringVar(&testFilterArgs.job, "job", "", "the name of the push or source job")
+		f.StringVar(&testFilterArgs.input, "input", "", "a filesystem name to test against the job's filters")
+		f.BoolVar(&testFilterArgs.all, "all", false, "test all local filesystems")
+	},
+	Run: runTestFilterCmd,
+}
+
+func runTestFilterCmd(subcommand *cli.Subcommand, args []string) error {
+
+	if testFilterArgs.job == "" {
+		return fmt.Errorf("must specify --job flag")
+	}
+	if !(testFilterArgs.all != (testFilterArgs.input != "")) { // xor
+		return fmt.Errorf("must set one: --all or --input")
+	}
+
+	conf := subcommand.Config()
+
+	var confFilter config.FilesystemsFilter
+	job, err := conf.Job(testFilterArgs.job)
+	if err != nil {
+		return err
+	}
+	switch j := job.Ret.(type) {
+	case *config.SourceJob: confFilter = j.Filesystems
+	case *config.PushJob: confFilter = j.Filesystems
+	default:
+		return fmt.Errorf("job type %T does not have filesystems filter", j)
+	}
+
+	f, err := filters.DatasetMapFilterFromConfig(confFilter)
+	if err != nil {
+		return fmt.Errorf("filter invalid: %s", err)
+	}
+
+	var fsnames []string
+	if testFilterArgs.input != "" {
+		fsnames = []string{testFilterArgs.input}
+	} else {
+		out, err := zfs.ZFSList([]string{"name"})
+		if err != nil {
+			return fmt.Errorf("could not list ZFS filesystems: %s", err)
+		}
+		for _, row := range out {
+
+			fsnames = append(fsnames, row[0])
+		}
+	}
+
+	fspaths := make([]*zfs.DatasetPath, len(fsnames))
+	for i, fsname := range fsnames {
+		path, err := zfs.NewDatasetPath(fsname)
+		if err != nil {
+			return err
+		}
+		fspaths[i] = path
+	}
+
+	hadFilterErr := false
+	for _, in := range fspaths {
+		var res string
+		var errStr string
+		pass, err := f.Filter(in)
+		if err != nil {
+			res = "ERROR"
+			errStr = err.Error()
+			hadFilterErr = true
+		} else if pass {
+			res = "ACCEPT"
+		} else {
+			res = "REJECT"
+		}
+		fmt.Printf("%s\t%s\t%s\n", res, in.ToString(), errStr)
+	}
+
+	if hadFilterErr {
+		return fmt.Errorf("filter errors occurred")
+	}
+	return nil
+}

client/version.go

@@ -0,0 +1,72 @@
+package client
+
+import (
+	"fmt"
+	"github.com/spf13/pflag"
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon"
+	"github.com/zrepl/zrepl/version"
+	"os"
+)
+
+var versionArgs struct {
+	Show   string
+	Config *config.Config
+	ConfigErr error
+}
+
+var VersionCmd = &cli.Subcommand{
+	Use:   "version",
+	Short: "print version of zrepl binary and running daemon",
+	NoRequireConfig: true,
+	SetupFlags: func(f *pflag.FlagSet) {
+		f.StringVar(&versionArgs.Show, "show", "", "version info to show (client|daemon)")
+	},
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		versionArgs.Config = subcommand.Config()
+		versionArgs.ConfigErr = subcommand.ConfigParsingError()
+		return runVersionCmd()
+	},
+}
+
+func runVersionCmd() error {
+	args := versionArgs
+
+	if args.Show != "daemon" && args.Show != "client" && args.Show != "" {
+		return fmt.Errorf("show flag must be 'client' or 'server' or be left empty")
+	}
+
+	var clientVersion, daemonVersion *version.ZreplVersionInformation
+	if args.Show == "client" || args.Show == "" {
+		clientVersion = version.NewZreplVersionInformation()
+		fmt.Printf("client: %s\n", clientVersion.String())
+	}
+	if args.Show == "daemon" || args.Show == "" {
+
+		if args.ConfigErr != nil {
+			return fmt.Errorf("config parsing error: %s", args.ConfigErr)
+		}
+
+		httpc, err := controlHttpClient(args.Config.Global.Control.SockPath)
+		if err != nil {
+			return fmt.Errorf("server: error: %s\n", err)
+		}
+
+		var info version.ZreplVersionInformation
+		err = jsonRequestResponse(httpc, daemon.ControlJobEndpointVersion, "", &info)
+		if err != nil {
+			return fmt.Errorf("server: error: %s\n", err)
+		}
+		daemonVersion = &info
+		fmt.Printf("server: %s\n", daemonVersion.String())
+	}
+
+	if args.Show == "" {
+		if clientVersion.Version != daemonVersion.Version {
+			fmt.Fprintf(os.Stderr, "WARNING: client version != daemon version, restart zrepl daemon\n")
+		}
+	}
+
+	return nil
+}

cmd/autosnap.go

@@ -1,91 +0,0 @@
-package cmd
-
-import (
-	"fmt"
-	"github.com/spf13/cobra"
-	"github.com/zrepl/zrepl/jobrun"
-	"github.com/zrepl/zrepl/zfs"
-	"os"
-	"sync"
-	"time"
-)
-
-var AutosnapCmd = &cobra.Command{
-	Use:   "autosnap",
-	Short: "perform automatic snapshotting",
-	Run:   cmdAutosnap,
-}
-
-func init() {
-	RootCmd.AddCommand(AutosnapCmd)
-}
-
-func cmdAutosnap(cmd *cobra.Command, args []string) {
-
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() {
-		defer wg.Done()
-		runner.Start()
-	}()
-
-	if len(args) < 1 {
-		log.Printf("must specify exactly one job as positional argument")
-		os.Exit(1)
-	}
-
-	snap, ok := conf.Autosnaps[args[0]]
-	if !ok {
-		log.Printf("could not find autosnap job: %s", args[0])
-		os.Exit(1)
-	}
-
-	job := jobrun.Job{
-		Name:           snap.JobName,
-		RepeatStrategy: snap.Interval,
-		RunFunc: func(log jobrun.Logger) error {
-			log.Printf("doing autosnap: %v", snap)
-			ctx := AutosnapContext{snap}
-			return doAutosnap(ctx, log)
-		},
-	}
-	runner.AddJob(job)
-
-	wg.Wait()
-
-}
-
-type AutosnapContext struct {
-	Autosnap *Autosnap
-}
-
-func doAutosnap(ctx AutosnapContext, log Logger) (err error) {
-
-	snap := ctx.Autosnap
-
-	filesystems, err := zfs.ZFSListMapping(snap.DatasetFilter)
-	if err != nil {
-		return fmt.Errorf("cannot filter datasets: %s", err)
-	}
-
-	suffix := time.Now().In(time.UTC).Format("20060102_150405_000")
-	snapname := fmt.Sprintf("%s%s", snap.Prefix, suffix)
-
-	hadError := false
-
-	for _, fs := range filesystems { // optimization: use recursive snapshots / channel programs here
-		log.Printf("snapshotting filesystem %s@%s", fs, snapname)
-		err := zfs.ZFSSnapshot(fs, snapname, false)
-		if err != nil {
-			log.Printf("error snapshotting %s: %s", fs, err)
-			hadError = true
-		}
-	}
-
-	if hadError {
-		err = fmt.Errorf("errors occurred during autosnap, check logs for details")
-	}
-
-	return
-
-}

cmd/config.go

@@ -1,730 +0,0 @@
-package cmd
-
-import (
-	"errors"
-	"fmt"
-	"github.com/jinzhu/copier"
-	"github.com/mitchellh/mapstructure"
-	"github.com/zrepl/zrepl/jobrun"
-	"github.com/zrepl/zrepl/rpc"
-	"github.com/zrepl/zrepl/sshbytestream"
-	. "github.com/zrepl/zrepl/util"
-	"github.com/zrepl/zrepl/zfs"
-	yaml "gopkg.in/yaml.v2"
-	"io"
-	"io/ioutil"
-	"regexp"
-	"strconv"
-	"strings"
-	"time"
-)
-
-var (
-	JobSectionPush     string = "push"
-	JobSectionPull     string = "pull"
-	JobSectionPrune    string = "prune"
-	JobSectionAutosnap string = "autosnap"
-)
-
-type Remote struct {
-	Name      string
-	Transport Transport
-}
-
-type Transport interface {
-	Connect(rpcLog Logger) (rpc.RPCRequester, error)
-}
-type LocalTransport struct {
-	Handler rpc.RPCHandler
-}
-type SSHTransport struct {
-	Host                 string
-	User                 string
-	Port                 uint16
-	IdentityFile         string   `mapstructure:"identity_file"`
-	TransportOpenCommand []string `mapstructure:"transport_open_command"`
-	SSHCommand           string   `mapstructure:"ssh_command"`
-	Options              []string
-	ConnLogReadFile      string `mapstructure:"connlog_read_file"`
-	ConnLogWriteFile     string `mapstructure:"connlog_write_file"`
-}
-
-type Push struct {
-	JobName           string // for use with jobrun package
-	To                *Remote
-	Filter            zfs.DatasetFilter
-	InitialReplPolicy rpc.InitialReplPolicy
-	RepeatStrategy    jobrun.RepeatStrategy
-}
-type Pull struct {
-	JobName           string // for use with jobrun package
-	From              *Remote
-	Mapping           DatasetMapFilter
-	InitialReplPolicy rpc.InitialReplPolicy
-	RepeatStrategy    jobrun.RepeatStrategy
-}
-
-type Prune struct {
-	JobName         string // for use with jobrun package
-	DatasetFilter   zfs.DatasetFilter
-	SnapshotFilter  zfs.FilesystemVersionFilter
-	RetentionPolicy *RetentionGrid // TODO abstract interface to support future policies?
-}
-
-type Autosnap struct {
-	JobName       string // for use with jobrun package
-	Prefix        string
-	Interval      jobrun.RepeatStrategy
-	DatasetFilter zfs.DatasetFilter
-}
-
-type Config struct {
-	Remotes   map[string]*Remote
-	Pushs     map[string]*Push            // job name -> job
-	Pulls     map[string]*Pull            // job name -> job
-	Sinks     map[string]DatasetMapFilter // client identity -> mapping
-	PullACLs  map[string]DatasetMapFilter // client identity -> filter
-	Prunes    map[string]*Prune           // job name -> job
-	Autosnaps map[string]*Autosnap        // job name -> job
-}
-
-func ParseConfig(path string) (config Config, err error) {
-
-	c := make(map[string]interface{}, 0)
-
-	var bytes []byte
-
-	if bytes, err = ioutil.ReadFile(path); err != nil {
-		return
-	}
-
-	if err = yaml.Unmarshal(bytes, &c); err != nil {
-		return
-	}
-
-	return parseMain(c)
-}
-
-func parseMain(root map[string]interface{}) (c Config, err error) {
-	if c.Remotes, err = parseRemotes(root["remotes"]); err != nil {
-		return
-	}
-
-	remoteLookup := func(name string) (remote *Remote, err error) {
-		remote = c.Remotes[name]
-		if remote == nil {
-			err = fmt.Errorf("remote '%s' not defined", name)
-		}
-		return
-	}
-
-	if c.Pushs, err = parsePushs(root["pushs"], remoteLookup); err != nil {
-		return
-	}
-	if c.Pulls, err = parsePulls(root["pulls"], remoteLookup); err != nil {
-		return
-	}
-	if c.Sinks, err = parseSinks(root["sinks"]); err != nil {
-		return
-	}
-	if c.PullACLs, err = parsePullACLs(root["pull_acls"]); err != nil {
-		return
-	}
-	if c.Prunes, err = parsePrunes(root["prune"]); err != nil {
-		return
-	}
-	if c.Autosnaps, err = parseAutosnaps(root["autosnap"]); err != nil {
-		return
-	}
-
-	return
-}
-
-func fullJobName(section, name string) (full string, err error) {
-	if len(name) < 1 {
-		err = fmt.Errorf("job name not set")
-		return
-	}
-	full = fmt.Sprintf("%s.%s", section, name)
-	return
-}
-
-func parseRemotes(v interface{}) (remotes map[string]*Remote, err error) {
-
-	asMap := make(map[string]struct {
-		Transport map[string]interface{}
-	}, 0)
-	if err = mapstructure.Decode(v, &asMap); err != nil {
-		return
-	}
-
-	remotes = make(map[string]*Remote, len(asMap))
-	for name, p := range asMap {
-
-		if name == rpc.LOCAL_TRANSPORT_IDENTITY {
-			err = errors.New(fmt.Sprintf("remote name '%s' reserved for local pulls", rpc.LOCAL_TRANSPORT_IDENTITY))
-			return
-		}
-
-		var transport Transport
-		if transport, err = parseTransport(p.Transport); err != nil {
-			return
-		}
-		remotes[name] = &Remote{
-			Name:      name,
-			Transport: transport,
-		}
-	}
-
-	return
-}
-
-func parseTransport(it map[string]interface{}) (t Transport, err error) {
-
-	if len(it) != 1 {
-		err = errors.New("ambiguous transport type")
-		return
-	}
-
-	for key, val := range it {
-		switch key {
-		case "ssh":
-			t := SSHTransport{}
-			if err = mapstructure.Decode(val, &t); err != nil {
-				err = errors.New(fmt.Sprintf("could not parse ssh transport: %s", err))
-				return nil, err
-			}
-			return t, nil
-		default:
-			return nil, errors.New(fmt.Sprintf("unknown transport type '%s'\n", key))
-		}
-	}
-
-	return // unreachable
-
-}
-
-type remoteLookup func(name string) (*Remote, error)
-
-func parsePushs(v interface{}, rl remoteLookup) (p map[string]*Push, err error) {
-
-	asMap := make(map[string]struct {
-		To                string
-		Filter            map[string]string
-		InitialReplPolicy string
-		Repeat            map[string]string
-	}, 0)
-
-	if err = mapstructure.Decode(v, &asMap); err != nil {
-		return
-	}
-
-	p = make(map[string]*Push, len(asMap))
-
-	for name, e := range asMap {
-
-		var toRemote *Remote
-		if toRemote, err = rl(e.To); err != nil {
-			return
-		}
-		push := &Push{
-			To: toRemote,
-		}
-
-		if push.JobName, err = fullJobName(JobSectionPush, name); err != nil {
-			return
-		}
-		if push.Filter, err = parseDatasetMapFilter(e.Filter, true); err != nil {
-			return
-		}
-
-		if push.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, rpc.DEFAULT_INITIAL_REPL_POLICY); err != nil {
-			return
-		}
-
-		if push.RepeatStrategy, err = parseRepeatStrategy(e.Repeat); err != nil {
-			return
-		}
-
-		p[name] = push
-	}
-
-	return
-}
-
-func parsePulls(v interface{}, rl remoteLookup) (p map[string]*Pull, err error) {
-
-	asMap := make(map[string]struct {
-		From              string
-		Mapping           map[string]string
-		InitialReplPolicy string
-		Repeat            map[string]string
-	}, 0)
-
-	if err = mapstructure.Decode(v, &asMap); err != nil {
-		return
-	}
-
-	p = make(map[string]*Pull, len(asMap))
-
-	for name, e := range asMap {
-
-		if len(e.From) < 1 {
-			err = fmt.Errorf("source not set ('from' attribute is empty)")
-			return
-		}
-
-		var fromRemote *Remote
-
-		if e.From == rpc.LOCAL_TRANSPORT_IDENTITY {
-			fromRemote = &Remote{
-				Name:      rpc.LOCAL_TRANSPORT_IDENTITY,
-				Transport: LocalTransport{},
-			}
-		} else {
-			if fromRemote, err = rl(e.From); err != nil {
-				return
-			}
-		}
-
-		pull := &Pull{
-			From: fromRemote,
-		}
-		if pull.JobName, err = fullJobName(JobSectionPull, name); err != nil {
-			return
-		}
-		if pull.Mapping, err = parseDatasetMapFilter(e.Mapping, false); err != nil {
-			return
-		}
-		if pull.InitialReplPolicy, err = parseInitialReplPolicy(e.InitialReplPolicy, rpc.DEFAULT_INITIAL_REPL_POLICY); err != nil {
-			return
-		}
-		if pull.RepeatStrategy, err = parseRepeatStrategy(e.Repeat); err != nil {
-			return
-		}
-
-		p[name] = pull
-	}
-
-	return
-}
-
-func parseInitialReplPolicy(v interface{}, defaultPolicy rpc.InitialReplPolicy) (p rpc.InitialReplPolicy, err error) {
-	s, ok := v.(string)
-	if !ok {
-		goto err
-	}
-
-	switch {
-	case s == "":
-		p = defaultPolicy
-	case s == "most_recent":
-		p = rpc.InitialReplPolicyMostRecent
-	case s == "all":
-		p = rpc.InitialReplPolicyAll
-	default:
-		goto err
-	}
-
-	return
-
-err:
-	err = errors.New(fmt.Sprintf("expected InitialReplPolicy, got %#v", v))
-	return
-}
-
-func parseRepeatStrategy(r map[string]string) (s jobrun.RepeatStrategy, err error) {
-
-	if r == nil {
-		return jobrun.NoRepeatStrategy{}, nil
-	}
-
-	if repeatStr, ok := r["interval"]; ok {
-		d, err := parseDuration(repeatStr)
-		if err != nil {
-			return nil, err
-		}
-		s = &jobrun.PeriodicRepeatStrategy{d}
-		return s, err
-	} else {
-		return nil, fmt.Errorf("attribute 'interval' not found but required in repeat specification")
-	}
-
-}
-
-func expectList(v interface{}) (asList []interface{}, err error) {
-	var ok bool
-	if asList, ok = v.([]interface{}); !ok {
-		err = errors.New("expected list")
-	}
-	return
-}
-
-func parseSinks(v interface{}) (m map[string]DatasetMapFilter, err error) {
-	var asMap map[string]map[string]interface{}
-	if err = mapstructure.Decode(v, &asMap); err != nil {
-		return
-	}
-
-	m = make(map[string]DatasetMapFilter, len(asMap))
-
-	for identity, entry := range asMap {
-		parseSink := func() (mapping DatasetMapFilter, err error) {
-			mappingMap, ok := entry["mapping"]
-			if !ok {
-				err = fmt.Errorf("no mapping specified")
-				return
-			}
-			mapping, err = parseDatasetMapFilter(mappingMap, false)
-			return
-		}
-		mapping, sinkErr := parseSink()
-		if sinkErr != nil {
-			err = fmt.Errorf("cannot parse sink for identity '%s': %s", identity, sinkErr)
-			return
-		}
-		m[identity] = mapping
-	}
-	return
-}
-
-func parsePullACLs(v interface{}) (m map[string]DatasetMapFilter, err error) {
-	var asMap map[string]map[string]interface{}
-	if err = mapstructure.Decode(v, &asMap); err != nil {
-		return
-	}
-
-	m = make(map[string]DatasetMapFilter, len(asMap))
-
-	for identity, entry := range asMap {
-		parsePullACL := func() (filter DatasetMapFilter, err error) {
-			filterMap, ok := entry["filter"]
-			if !ok {
-				err = fmt.Errorf("no filter specified")
-				return
-			}
-			filter, err = parseDatasetMapFilter(filterMap, true)
-			return
-		}
-		filter, filterErr := parsePullACL()
-		if filterErr != nil {
-			err = fmt.Errorf("cannot parse pull-ACL for identity '%s': %s", identity, filterErr)
-			return
-		}
-		m[identity] = filter
-	}
-	return
-}
-
-func parseDatasetMapFilter(mi interface{}, filterOnly bool) (f DatasetMapFilter, err error) {
-
-	var m map[string]string
-	if err = mapstructure.Decode(mi, &m); err != nil {
-		err = fmt.Errorf("maps / filters must be specified as map[string]string: %s", err)
-		return
-	}
-
-	f = NewDatasetMapFilter(len(m), filterOnly)
-	for pathPattern, mapping := range m {
-		if err = f.Add(pathPattern, mapping); err != nil {
-			err = fmt.Errorf("invalid mapping entry ['%s':'%s']: %s", pathPattern, mapping, err)
-			return
-		}
-	}
-	return
-}
-
-func (t SSHTransport) Connect(rpcLog Logger) (r rpc.RPCRequester, err error) {
-	var stream io.ReadWriteCloser
-	var rpcTransport sshbytestream.SSHTransport
-	if err = copier.Copy(&rpcTransport, t); err != nil {
-		return
-	}
-	if stream, err = sshbytestream.Outgoing(rpcTransport); err != nil {
-		return
-	}
-	stream, err = NewReadWriteCloserLogger(stream, t.ConnLogReadFile, t.ConnLogWriteFile)
-	if err != nil {
-		return
-	}
-	return rpc.ConnectByteStreamRPC(stream, rpcLog)
-}
-
-func (t LocalTransport) Connect(rpcLog Logger) (r rpc.RPCRequester, err error) {
-	if t.Handler == nil {
-		panic("local transport with uninitialized handler")
-	}
-	return rpc.ConnectLocalRPC(t.Handler), nil
-}
-
-func (t *LocalTransport) SetHandler(handler rpc.RPCHandler) {
-	t.Handler = handler
-}
-
-func parsePrunes(m interface{}) (rets map[string]*Prune, err error) {
-
-	asList := make(map[string]map[string]interface{}, 0)
-	if err = mapstructure.Decode(m, &asList); err != nil {
-		return
-	}
-
-	rets = make(map[string]*Prune, len(asList))
-
-	for name, e := range asList {
-		var prune *Prune
-		if prune, err = parsePrune(e, name); err != nil {
-			err = fmt.Errorf("cannot parse prune job %s: %s", name, err)
-			return
-		}
-		rets[name] = prune
-	}
-
-	return
-}
-
-func parsePrune(e map[string]interface{}, name string) (prune *Prune, err error) {
-
-	// Only support grid policy for now
-	policyName, ok := e["policy"]
-	if !ok || policyName != "grid" {
-		err = fmt.Errorf("prune job with unimplemented policy '%s'", policyName)
-		return
-	}
-
-	var i struct {
-		Grid           string
-		DatasetFilter  map[string]string `mapstructure:"dataset_filter"`
-		SnapshotFilter map[string]string `mapstructure:"snapshot_filter"`
-	}
-
-	if err = mapstructure.Decode(e, &i); err != nil {
-		return
-	}
-
-	prune = &Prune{}
-
-	if prune.JobName, err = fullJobName(JobSectionPrune, name); err != nil {
-		return
-	}
-
-	// Parse grid policy
-	intervals, err := parseRetentionGridIntervalsString(i.Grid)
-	if err != nil {
-		err = fmt.Errorf("cannot parse retention grid: %s", err)
-		return
-	}
-	// Assert intervals are of increasing length (not necessarily required, but indicates config mistake)
-	lastDuration := time.Duration(0)
-	for i := range intervals {
-		if intervals[i].Length < lastDuration {
-			err = fmt.Errorf("retention grid interval length must be monotonically increasing:"+
-				"interval %d is shorter than %d", i+1, i)
-			return
-		} else {
-			lastDuration = intervals[i].Length
-		}
-	}
-	prune.RetentionPolicy = NewRetentionGrid(intervals)
-
-	// Parse filters
-	if prune.DatasetFilter, err = parseDatasetMapFilter(i.DatasetFilter, true); err != nil {
-		err = fmt.Errorf("cannot parse dataset filter: %s", err)
-		return
-	}
-	if prune.SnapshotFilter, err = parseSnapshotFilter(i.SnapshotFilter); err != nil {
-		err = fmt.Errorf("cannot parse snapshot filter: %s", err)
-		return
-	}
-
-	return
-}
-
-var durationStringRegex *regexp.Regexp = regexp.MustCompile(`^\s*(\d+)\s*(s|m|h|d|w)\s*$`)
-
-func parseDuration(e string) (d time.Duration, err error) {
-	comps := durationStringRegex.FindStringSubmatch(e)
-	if len(comps) != 3 {
-		err = fmt.Errorf("does not match regex: %s %#v", e, comps)
-		return
-	}
-
-	durationFactor, err := strconv.ParseInt(comps[1], 10, 64)
-	if err != nil {
-		return
-	}
-
-	var durationUnit time.Duration
-	switch comps[2] {
-	case "s":
-		durationUnit = time.Second
-	case "m":
-		durationUnit = time.Minute
-	case "h":
-		durationUnit = time.Hour
-	case "d":
-		durationUnit = 24 * time.Hour
-	case "w":
-		durationUnit = 24 * 7 * time.Hour
-	default:
-		err = fmt.Errorf("contains unknown time unit '%s'", comps[2])
-		return
-	}
-
-	d = time.Duration(durationFactor) * durationUnit
-	return
-
-}
-
-var retentionStringIntervalRegex *regexp.Regexp = regexp.MustCompile(`^\s*(\d+)\s*x\s*([^\(]+)\s*(\((.*)\))?\s*$`)
-
-func parseRetentionGridIntervalString(e string) (intervals []RetentionInterval, err error) {
-
-	comps := retentionStringIntervalRegex.FindStringSubmatch(e)
-	if comps == nil {
-		err = fmt.Errorf("retention string does not match expected format")
-		return
-	}
-
-	times, err := strconv.Atoi(comps[1])
-	if err != nil {
-		return nil, err
-	} else if times <= 0 {
-		return nil, fmt.Errorf("contains factor <= 0")
-	}
-
-	duration, err := parseDuration(comps[2])
-	if err != nil {
-		return nil, err
-	}
-
-	keepCount := 1
-	if comps[3] != "" {
-		// Decompose key=value, comma separated
-		// For now, only keep_count is supported
-		re := regexp.MustCompile(`^\s*keep=(.+)\s*$`)
-		res := re.FindStringSubmatch(comps[4])
-		if res == nil || len(res) != 2 {
-			err = fmt.Errorf("interval parameter contains unknown parameters")
-			return
-		}
-		if res[1] == "all" {
-			keepCount = RetentionGridKeepCountAll
-		} else {
-			keepCount, err = strconv.Atoi(res[1])
-			if err != nil {
-				err = fmt.Errorf("cannot parse keep_count value")
-				return
-			}
-		}
-	}
-
-	intervals = make([]RetentionInterval, times)
-	for i := range intervals {
-		intervals[i] = RetentionInterval{
-			Length:    duration,
-			KeepCount: keepCount,
-		}
-	}
-
-	return
-
-}
-
-func parseRetentionGridIntervalsString(s string) (intervals []RetentionInterval, err error) {
-
-	ges := strings.Split(s, "|")
-	intervals = make([]RetentionInterval, 0, 7*len(ges))
-
-	for intervalIdx, e := range ges {
-		parsed, err := parseRetentionGridIntervalString(e)
-		if err != nil {
-			return nil, fmt.Errorf("cannot parse interval %d of %d: %s: %s", intervalIdx+1, len(ges), err, strings.TrimSpace(e))
-		}
-		intervals = append(intervals, parsed...)
-	}
-
-	return
-}
-
-type prefixSnapshotFilter struct {
-	prefix string
-}
-
-func (f prefixSnapshotFilter) Filter(fsv zfs.FilesystemVersion) (accept bool, err error) {
-	return fsv.Type == zfs.Snapshot && strings.HasPrefix(fsv.Name, f.prefix), nil
-}
-
-func parseSnapshotFilter(fm map[string]string) (snapFilter zfs.FilesystemVersionFilter, err error) {
-	prefix, ok := fm["prefix"]
-	if !ok {
-		err = fmt.Errorf("unsupported snapshot filter")
-		return
-	}
-	snapFilter = prefixSnapshotFilter{prefix}
-	return
-}
-
-func parseAutosnaps(m interface{}) (snaps map[string]*Autosnap, err error) {
-
-	asMap := make(map[string]interface{}, 0)
-	if err = mapstructure.Decode(m, &asMap); err != nil {
-		return
-	}
-
-	snaps = make(map[string]*Autosnap, len(asMap))
-
-	for name, e := range asMap {
-		var snap *Autosnap
-		if snap, err = parseAutosnap(e, name); err != nil {
-			err = fmt.Errorf("cannot parse autonsap job %s: %s", name, err)
-			return
-		}
-		snaps[name] = snap
-	}
-
-	return
-
-}
-
-func parseAutosnap(m interface{}, name string) (a *Autosnap, err error) {
-
-	var i struct {
-		Prefix        string
-		Interval      string
-		DatasetFilter map[string]string `mapstructure:"dataset_filter"`
-	}
-
-	if err = mapstructure.Decode(m, &i); err != nil {
-		err = fmt.Errorf("structure unfit: %s", err)
-		return
-	}
-
-	a = &Autosnap{}
-
-	if a.JobName, err = fullJobName(JobSectionAutosnap, name); err != nil {
-		return
-	}
-
-	if len(i.Prefix) < 1 {
-		err = fmt.Errorf("prefix must not be empty")
-		return
-	}
-	a.Prefix = i.Prefix
-
-	var interval time.Duration
-	if interval, err = parseDuration(i.Interval); err != nil {
-		err = fmt.Errorf("cannot parse interval: %s", err)
-		return
-	}
-	a.Interval = &jobrun.PeriodicRepeatStrategy{interval}
-
-	if len(i.DatasetFilter) == 0 {
-		err = fmt.Errorf("dataset_filter not specified")
-		return
-	}
-	if a.DatasetFilter, err = parseDatasetMapFilter(i.DatasetFilter, true); err != nil {
-		err = fmt.Errorf("cannot parse dataset filter: %s", err)
-	}
-
-	return
-}

cmd/config_mapfilter.go

@@ -1,162 +0,0 @@
-package cmd
-
-import (
-	"errors"
-	"fmt"
-	"strings"
-
-	"github.com/zrepl/zrepl/zfs"
-)
-
-type DatasetMapFilter struct {
-	entries []datasetMapFilterEntry
-
-	// if set, only valid filter entries can be added using Add()
-	// and Map() will always return an error
-	filterOnly bool
-}
-
-type datasetMapFilterEntry struct {
-	path *zfs.DatasetPath
-	// the mapping. since this datastructure acts as both mapping and filter
-	// we have to convert it to the desired rep dynamically
-	mapping      string
-	subtreeMatch bool
-}
-
-var NoMatchError error = errors.New("no match found in mapping")
-
-func NewDatasetMapFilter(capacity int, filterOnly bool) DatasetMapFilter {
-	return DatasetMapFilter{
-		entries: make([]datasetMapFilterEntry, 0, capacity),
-	}
-}
-
-func (m *DatasetMapFilter) Add(pathPattern, mapping string) (err error) {
-
-	if m.filterOnly {
-		if _, err = parseDatasetFilterResult(mapping); err != nil {
-			return
-		}
-	}
-
-	// assert path glob adheres to spec
-	const SUBTREE_PATTERN string = "<"
-	patternCount := strings.Count(pathPattern, SUBTREE_PATTERN)
-	switch {
-	case patternCount > 1:
-	case patternCount == 1 && !strings.HasSuffix(pathPattern, SUBTREE_PATTERN):
-		err = fmt.Errorf("pattern invalid: only one '<' at end of string allowed")
-		return
-	}
-
-	pathStr := strings.TrimSuffix(pathPattern, SUBTREE_PATTERN)
-	path, err := zfs.NewDatasetPath(pathStr)
-	if err != nil {
-		return fmt.Errorf("pattern is not a dataset path: %s", err)
-	}
-
-	entry := datasetMapFilterEntry{
-		path:         path,
-		mapping:      mapping,
-		subtreeMatch: patternCount > 0,
-	}
-	m.entries = append(m.entries, entry)
-	return
-
-}
-
-// find the most specific prefix mapping we have
-//
-// longer prefix wins over shorter prefix, direct wins over glob
-func (m DatasetMapFilter) mostSpecificPrefixMapping(path *zfs.DatasetPath) (idx int, found bool) {
-	lcp, lcp_entry_idx := -1, -1
-	direct_idx := -1
-	for e := range m.entries {
-		entry := m.entries[e]
-		ep := m.entries[e].path
-		lep := ep.Length()
-
-		switch {
-		case !entry.subtreeMatch && ep.Equal(path):
-			direct_idx = e
-			continue
-		case entry.subtreeMatch && path.HasPrefix(ep) && lep > lcp:
-			lcp = lep
-			lcp_entry_idx = e
-		default:
-			continue
-		}
-	}
-
-	if lcp_entry_idx >= 0 || direct_idx >= 0 {
-		found = true
-		switch {
-		case direct_idx >= 0:
-			idx = direct_idx
-		case lcp_entry_idx >= 0:
-			idx = lcp_entry_idx
-		}
-	}
-	return
-}
-
-func (m DatasetMapFilter) Map(source *zfs.DatasetPath) (target *zfs.DatasetPath, err error) {
-
-	if m.filterOnly {
-		err = fmt.Errorf("using a filter for mapping simply does not work")
-		return
-	}
-
-	mi, hasMapping := m.mostSpecificPrefixMapping(source)
-	if !hasMapping {
-		err = NoMatchError
-		return
-	}
-	me := m.entries[mi]
-
-	target, err = zfs.NewDatasetPath(me.mapping)
-	if err != nil {
-		err = fmt.Errorf("mapping target is not a dataset path: %s", err)
-		return
-	}
-	if m.entries[mi].subtreeMatch {
-		// strip common prefix
-		extendComps := source.Copy()
-		if me.path.Empty() {
-			// special case: trying to map the root => strip first component
-			extendComps.TrimNPrefixComps(1)
-		} else {
-			extendComps.TrimPrefix(me.path)
-		}
-		target.Extend(extendComps)
-	}
-	return
-}
-
-func (m DatasetMapFilter) Filter(p *zfs.DatasetPath) (pass bool, err error) {
-	mi, hasMapping := m.mostSpecificPrefixMapping(p)
-	if !hasMapping {
-		pass = false
-		return
-	}
-	me := m.entries[mi]
-	pass, err = parseDatasetFilterResult(me.mapping)
-	return
-}
-
-// Parse a dataset filter result
-func parseDatasetFilterResult(result string) (pass bool, err error) {
-	l := strings.ToLower(result)
-	switch strings.ToLower(l) {
-	case "ok":
-		pass = true
-		return
-	case "omit":
-		return
-	default:
-		err = fmt.Errorf("'%s' is not a valid filter result", result)
-		return
-	}
-	return
-}

cmd/config_test.go

@@ -1,123 +0,0 @@
-package cmd
-
-import (
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/zrepl/zrepl/util"
-	"github.com/zrepl/zrepl/zfs"
-)
-
-func TestSampleConfigFileIsParsedWithoutErrors(t *testing.T) {
-	_, err := ParseConfig("./sampleconf/zrepl.yml")
-	assert.Nil(t, err)
-}
-
-func TestParseRetentionGridStringParsing(t *testing.T) {
-
-	intervals, err := parseRetentionGridIntervalsString("2x10m(keep=2) | 1x1h | 3x1w")
-
-	assert.Nil(t, err)
-	assert.Len(t, intervals, 6)
-	proto := util.RetentionInterval{
-		KeepCount: 2,
-		Length:    10 * time.Minute,
-	}
-	assert.EqualValues(t, proto, intervals[0])
-	assert.EqualValues(t, proto, intervals[1])
-
-	proto.KeepCount = 1
-	proto.Length = 1 * time.Hour
-	assert.EqualValues(t, proto, intervals[2])
-
-	proto.Length = 7 * 24 * time.Hour
-	assert.EqualValues(t, proto, intervals[3])
-	assert.EqualValues(t, proto, intervals[4])
-	assert.EqualValues(t, proto, intervals[5])
-
-	intervals, err = parseRetentionGridIntervalsString("|")
-	assert.Error(t, err)
-	intervals, err = parseRetentionGridIntervalsString("2x10m")
-	assert.NoError(t, err)
-
-	intervals, err = parseRetentionGridIntervalsString("1x10m(keep=all)")
-	assert.NoError(t, err)
-	assert.Len(t, intervals, 1)
-	assert.EqualValues(t, util.RetentionGridKeepCountAll, intervals[0].KeepCount)
-
-}
-
-func TestDatasetMapFilter(t *testing.T) {
-
-	expectMapping := func(m map[string]string, from, to string) {
-		dmf, err := parseDatasetMapFilter(m, false)
-		if err != nil {
-			t.Logf("expect test map to be valid: %s", err)
-			t.FailNow()
-		}
-		fromPath, err := zfs.NewDatasetPath(from)
-		if err != nil {
-			t.Logf("expect test from path to be valid: %s", err)
-			t.FailNow()
-		}
-		toPath, err := zfs.NewDatasetPath(to)
-		if err != nil {
-			t.Logf("expect test to path to be valid: %s", err)
-			t.FailNow()
-		}
-
-		res, err := dmf.Map(fromPath)
-		t.Logf("%s => %s", fromPath.ToString(), res.ToString())
-		assert.Nil(t, err)
-		assert.True(t, res.Equal(toPath))
-	}
-
-	expectFilter := func(m map[string]string, path string, pass bool) {
-		dmf, err := parseDatasetMapFilter(m, true)
-		if err != nil {
-			t.Logf("expect test filter to be valid: %s", err)
-			t.FailNow()
-		}
-		p, err := zfs.NewDatasetPath(path)
-		if err != nil {
-			t.Logf("expect test path to be valid: %s", err)
-			t.FailNow()
-		}
-		res, err := dmf.Filter(p)
-		assert.Nil(t, err)
-		assert.Equal(t, pass, res)
-	}
-
-	map1 := map[string]string{
-		"a/b/c<": "root1",
-		"a/b<":   "root2",
-		"<":      "root3/b/c",
-		"q<":     "root4/1/2",
-	}
-
-	expectMapping(map1, "a/b/c", "root1")
-	expectMapping(map1, "a/b/c/d", "root1/d")
-	expectMapping(map1, "a/b/e", "root2/e")
-	expectMapping(map1, "a/b", "root2")
-	expectMapping(map1, "x", "root3/b/c")
-	expectMapping(map1, "x/y", "root3/b/c/y")
-	expectMapping(map1, "q", "root4/1/2")
-	expectMapping(map1, "q/r", "root4/1/2/r")
-
-	filter1 := map[string]string{
-		"<":    "omit",
-		"a<":   "ok",
-		"a/b<": "omit",
-	}
-
-	expectFilter(filter1, "b", false)
-	expectFilter(filter1, "a", true)
-	expectFilter(filter1, "a/d", true)
-	expectFilter(filter1, "a/b", false)
-	expectFilter(filter1, "a/b/c", false)
-
-	filter2 := map[string]string{}
-	expectFilter(filter2, "foo", false) // default to omit
-
-}

cmd/handler.go

@@ -1,137 +0,0 @@
-package cmd
-
-import (
-	"fmt"
-	"github.com/zrepl/zrepl/rpc"
-	"github.com/zrepl/zrepl/zfs"
-	"io"
-)
-
-type DatasetMapping interface {
-	Map(source *zfs.DatasetPath) (target *zfs.DatasetPath, err error)
-}
-
-type Handler struct {
-	Logger          Logger
-	PullACL         zfs.DatasetFilter
-	SinkMappingFunc func(clientIdentity string) (mapping DatasetMapping, err error)
-}
-
-func (h Handler) HandleFilesystemRequest(r rpc.FilesystemRequest) (roots []*zfs.DatasetPath, err error) {
-
-	h.Logger.Printf("handling fsr: %#v", r)
-
-	h.Logger.Printf("using PullACL: %#v", h.PullACL)
-
-	if roots, err = zfs.ZFSListMapping(h.PullACL); err != nil {
-		h.Logger.Printf("handle fsr err: %v\n", err)
-		return
-	}
-
-	h.Logger.Printf("returning: %#v", roots)
-
-	return
-}
-
-func (h Handler) HandleFilesystemVersionsRequest(r rpc.FilesystemVersionsRequest) (versions []zfs.FilesystemVersion, err error) {
-
-	h.Logger.Printf("handling filesystem versions request: %#v", r)
-
-	// allowed to request that?
-	if h.pullACLCheck(r.Filesystem); err != nil {
-		return
-	}
-
-	// find our versions
-	if versions, err = zfs.ZFSListFilesystemVersions(r.Filesystem, nil); err != nil {
-		h.Logger.Printf("our versions error: %#v\n", err)
-		return
-	}
-
-	h.Logger.Printf("our versions: %#v\n", versions)
-	return
-
-}
-
-func (h Handler) HandleInitialTransferRequest(r rpc.InitialTransferRequest) (stream io.Reader, err error) {
-
-	h.Logger.Printf("handling initial transfer request: %#v", r)
-	if err = h.pullACLCheck(r.Filesystem); err != nil {
-		return
-	}
-
-	h.Logger.Printf("invoking zfs send")
-
-	if stream, err = zfs.ZFSSend(r.Filesystem, &r.FilesystemVersion, nil); err != nil {
-		h.Logger.Printf("error sending filesystem: %#v", err)
-	}
-
-	return
-
-}
-
-func (h Handler) HandleIncrementalTransferRequest(r rpc.IncrementalTransferRequest) (stream io.Reader, err error) {
-
-	h.Logger.Printf("handling incremental transfer request: %#v", r)
-	if err = h.pullACLCheck(r.Filesystem); err != nil {
-		return
-	}
-
-	h.Logger.Printf("invoking zfs send")
-
-	if stream, err = zfs.ZFSSend(r.Filesystem, &r.From, &r.To); err != nil {
-		h.Logger.Printf("error sending filesystem: %#v", err)
-	}
-
-	return
-
-}
-
-func (h Handler) HandlePullMeRequest(r rpc.PullMeRequest, clientIdentity string, client rpc.RPCRequester) (err error) {
-
-	// Check if we have a sink for this request
-	// Use that mapping to do what happens in doPull
-
-	h.Logger.Printf("handling PullMeRequest: %#v", r)
-
-	var sinkMapping DatasetMapping
-	sinkMapping, err = h.SinkMappingFunc(clientIdentity)
-	if err != nil {
-		h.Logger.Printf("no sink mapping for client identity '%s', denying PullMeRequest", clientIdentity)
-		err = fmt.Errorf("no sink for client identity '%s'", clientIdentity)
-		return
-	}
-
-	h.Logger.Printf("doing pull...")
-
-	err = doPull(PullContext{
-		Remote:            client,
-		Log:               h.Logger,
-		Mapping:           sinkMapping,
-		InitialReplPolicy: r.InitialReplPolicy,
-	})
-	if err != nil {
-		h.Logger.Printf("PullMeRequest failed with error: %s", err)
-		return
-	}
-
-	h.Logger.Printf("finished handling PullMeRequest: %#v", r)
-
-	return
-}
-
-func (h Handler) pullACLCheck(p *zfs.DatasetPath) (err error) {
-	var allowed bool
-	allowed, err = h.PullACL.Filter(p)
-	if err != nil {
-		err = fmt.Errorf("error evaluating ACL: %s", err)
-		h.Logger.Printf(err.Error())
-		return
-	}
-	if !allowed {
-		err = fmt.Errorf("ACL prohibits access to %s", p.ToString())
-		h.Logger.Printf(err.Error())
-		return
-	}
-	return
-}

cmd/main.go

@@ -1,104 +0,0 @@
-// zrepl replicates ZFS filesystems & volumes between pools
-//
-// Code Organization
-//
-// The cmd package uses github.com/spf13/cobra for its CLI.
-//
-// It combines the other packages in the zrepl project to implement zrepl functionality.
-//
-// Each subcommand's code is in the corresponding *.go file.
-// All other *.go files contain code shared by the subcommands.
-package cmd
-
-import (
-	"fmt"
-	"github.com/spf13/cobra"
-	"github.com/zrepl/zrepl/jobrun"
-	"golang.org/x/sys/unix"
-	"io"
-	golog "log"
-	"net/http"
-	_ "net/http/pprof"
-	"os"
-)
-
-type Logger interface {
-	Printf(format string, v ...interface{})
-}
-
-// global state / facilities
-var (
-	conf     Config
-	runner   *jobrun.JobRunner
-	logFlags int = golog.LUTC | golog.Ldate | golog.Ltime
-	logOut   io.Writer
-	log      Logger
-)
-
-var RootCmd = &cobra.Command{
-	Use:   "zrepl",
-	Short: "ZFS dataset replication",
-	Long: `Replicate ZFS filesystems & volumes between pools:
-
-  - push & pull mode
-  - automatic snapshot creation & pruning
-  - local / over the network
-  - ACLs instead of blank SSH access`,
-}
-
-var rootArgs struct {
-	configFile string
-	stderrFile string
-	httpPprof  string
-}
-
-func init() {
-	cobra.OnInitialize(initConfig)
-	RootCmd.PersistentFlags().StringVar(&rootArgs.configFile, "config", "", "config file path")
-	RootCmd.PersistentFlags().StringVar(&rootArgs.stderrFile, "stderrFile", "-", "redirect stderr to given path")
-	RootCmd.PersistentFlags().StringVar(&rootArgs.httpPprof, "debug.pprof.http", "", "run pprof http server on given port")
-}
-
-func initConfig() {
-
-	// Logging & stderr redirection
-	if rootArgs.stderrFile != "-" {
-		file, err := os.OpenFile(rootArgs.stderrFile, os.O_CREATE|os.O_TRUNC|os.O_WRONLY, 0600)
-		if err != nil {
-			return
-		}
-
-		if err = unix.Dup2(int(file.Fd()), int(os.Stderr.Fd())); err != nil {
-			file.WriteString(fmt.Sprintf("error redirecting stderr file %s: %s\n", rootArgs.stderrFile, err))
-			return
-		}
-		logOut = file
-	} else {
-		logOut = os.Stderr
-	}
-
-	log = golog.New(logOut, "", logFlags)
-
-	// CPU profiling
-	if rootArgs.httpPprof != "" {
-		go func() {
-			http.ListenAndServe(rootArgs.httpPprof, nil)
-		}()
-	}
-
-	// Config
-	if rootArgs.configFile == "" {
-		log.Printf("config file not set")
-		os.Exit(1)
-	}
-	var err error
-	if conf, err = ParseConfig(rootArgs.configFile); err != nil {
-		log.Printf("error parsing config: %s", err)
-		os.Exit(1)
-	}
-
-	jobrunLogger := golog.New(os.Stderr, "jobrun ", logFlags)
-	runner = jobrun.NewJobRunner(jobrunLogger)
-	return
-
-}

cmd/prune.go

@@ -1,139 +0,0 @@
-package cmd
-
-import (
-	"fmt"
-	"github.com/spf13/cobra"
-	"github.com/zrepl/zrepl/util"
-	"github.com/zrepl/zrepl/zfs"
-	"os"
-	"sort"
-	"time"
-)
-
-var pruneArgs struct {
-	job    string
-	dryRun bool
-}
-
-var PruneCmd = &cobra.Command{
-	Use:   "prune",
-	Short: "perform pruning",
-	Run:   cmdPrune,
-}
-
-func init() {
-	PruneCmd.Flags().StringVar(&pruneArgs.job, "job", "", "job to run")
-	PruneCmd.Flags().BoolVarP(&pruneArgs.dryRun, "dryrun", "n", false, "dry run")
-	RootCmd.AddCommand(PruneCmd)
-}
-
-func cmdPrune(cmd *cobra.Command, args []string) {
-
-	if len(args) < 1 {
-		log.Printf("must specify exactly one job as positional argument")
-		os.Exit(1)
-	}
-
-	job, ok := conf.Prunes[args[0]]
-	if !ok {
-		log.Printf("could not find prune job: %s", args[0])
-		os.Exit(1)
-	}
-
-	log.Printf("Beginning prune job:\n%s", job)
-	ctx := PruneContext{job, time.Now(), pruneArgs.dryRun}
-	err := doPrune(ctx, log)
-	if err != nil {
-		log.Printf("Prune job failed with error: %s", err)
-		os.Exit(1)
-	}
-
-}
-
-type PruneContext struct {
-	Prune  *Prune
-	Now    time.Time
-	DryRun bool
-}
-
-type retentionGridAdaptor struct {
-	zfs.FilesystemVersion
-}
-
-func (a retentionGridAdaptor) Date() time.Time {
-	return a.Creation
-}
-
-func (a retentionGridAdaptor) LessThan(b util.RetentionGridEntry) bool {
-	return a.CreateTXG < b.(retentionGridAdaptor).CreateTXG
-}
-
-func doPrune(ctx PruneContext, log Logger) error {
-
-	if ctx.DryRun {
-		log.Printf("doing dry run")
-	}
-
-	prune := ctx.Prune
-
-	// ZFSListSnapsFiltered --> todo can extend fsfilter or need new? Have already something per fs
-	// Dedicated snapshot object? Adaptor object to FilesystemVersion?
-
-	filesystems, err := zfs.ZFSListMapping(prune.DatasetFilter)
-	if err != nil {
-		return fmt.Errorf("error applying filesystem filter: %s", err)
-	}
-
-	for _, fs := range filesystems {
-
-		fsversions, err := zfs.ZFSListFilesystemVersions(fs, prune.SnapshotFilter)
-		if err != nil {
-			return fmt.Errorf("error listing filesytem versions of %s: %s", fs, err)
-		}
-
-		if len(fsversions) == 0 {
-			continue
-		}
-
-		adaptors := make([]util.RetentionGridEntry, len(fsversions))
-		for fsv := range fsversions {
-			adaptors[fsv] = retentionGridAdaptor{fsversions[fsv]}
-		}
-
-		sort.SliceStable(adaptors, func(i, j int) bool {
-			return adaptors[i].LessThan(adaptors[j])
-		})
-		ctx.Now = adaptors[len(adaptors)-1].Date()
-
-		describe := func(a retentionGridAdaptor) string {
-			timeSince := a.Creation.Sub(ctx.Now)
-			const day time.Duration = 24 * time.Hour
-			days := timeSince / day
-			remainder := timeSince % day
-			return fmt.Sprintf("%s@%dd%s from now", a.ToAbsPath(fs), days, remainder)
-		}
-
-		keep, remove := prune.RetentionPolicy.FitEntries(ctx.Now, adaptors)
-		for _, a := range remove {
-			r := a.(retentionGridAdaptor)
-			log.Printf("remove %s", describe(r))
-			// do echo what we'll do and exec zfs destroy if not dry run
-			// special handling for EBUSY (zfs hold)
-			// error handling for clones? just echo to cli, skip over, and exit with non-zero status code (we're idempotent)
-			if !ctx.DryRun {
-				err := zfs.ZFSDestroy(r.ToAbsPath(fs))
-				if err != nil {
-					// handle
-					log.Printf("error: %s", err)
-				}
-			}
-		}
-		for _, a := range keep {
-			r := a.(retentionGridAdaptor)
-			log.Printf("would keep %s", describe(r))
-		}
-	}
-
-	return nil
-
-}

cmd/replication.go

@@ -1,506 +0,0 @@
-package cmd
-
-import (
-	"fmt"
-	"io"
-	"os"
-	"sync"
-	"time"
-
-	"github.com/spf13/cobra"
-	"github.com/zrepl/zrepl/jobrun"
-	"github.com/zrepl/zrepl/rpc"
-	"github.com/zrepl/zrepl/util"
-	"github.com/zrepl/zrepl/zfs"
-)
-
-var runArgs struct {
-	job  string
-	once bool
-}
-
-var RunCmd = &cobra.Command{
-	Use:   "run",
-	Short: "run push & pull replication",
-	Run:   cmdRun,
-}
-
-var PushCmd = &cobra.Command{
-	Use:   "push",
-	Short: "run push job (first positional argument)",
-	Run:   cmdPush,
-}
-
-var PullCmd = &cobra.Command{
-	Use:   "pull",
-	Short: "run pull job (first positional argument)",
-	Run:   cmdPull,
-}
-
-func init() {
-	RootCmd.AddCommand(RunCmd)
-	RunCmd.Flags().BoolVar(&runArgs.once, "once", false, "run jobs only once, regardless of configured repeat behavior")
-	RunCmd.Flags().StringVar(&runArgs.job, "job", "", "run only the given job")
-
-	RootCmd.AddCommand(PushCmd)
-	RootCmd.AddCommand(PullCmd)
-}
-
-func cmdPush(cmd *cobra.Command, args []string) {
-
-	if len(args) != 1 {
-		log.Printf("must specify exactly one job as positional argument")
-		os.Exit(1)
-	}
-	job, ok := conf.Pushs[args[0]]
-	if !ok {
-		log.Printf("could not find push job %s", args[0])
-		os.Exit(1)
-	}
-	if err := jobPush(job, log); err != nil {
-		log.Printf("error doing push: %s", err)
-		os.Exit(1)
-	}
-
-}
-
-func cmdPull(cmd *cobra.Command, args []string) {
-
-	if len(args) != 1 {
-		log.Printf("must specify exactly one job as positional argument")
-		os.Exit(1)
-	}
-	job, ok := conf.Pulls[args[0]]
-	if !ok {
-		log.Printf("could not find pull job %s", args[0])
-		os.Exit(1)
-	}
-
-	if err := jobPull(job, log); err != nil {
-		log.Printf("error doing pull: %s", err)
-		os.Exit(1)
-	}
-
-}
-
-func cmdRun(cmd *cobra.Command, args []string) {
-
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() {
-		defer wg.Done()
-		runner.Start()
-	}()
-
-	jobs := make([]jobrun.Job, len(conf.Pulls)+len(conf.Pushs))
-	i := 0
-	for _, pull := range conf.Pulls {
-		jobs[i] = jobrun.Job{
-			Name:           fmt.Sprintf("pull.%d", i),
-			RepeatStrategy: pull.RepeatStrategy,
-			RunFunc: func(log jobrun.Logger) error {
-				log.Printf("doing pull: %v", pull)
-				return jobPull(pull, log)
-			},
-		}
-		i++
-	}
-	for _, push := range conf.Pushs {
-		jobs[i] = jobrun.Job{
-			Name:           fmt.Sprintf("push.%d", i),
-			RepeatStrategy: push.RepeatStrategy,
-			RunFunc: func(log jobrun.Logger) error {
-				log.Printf("doing push: %v", push)
-				return jobPush(push, log)
-			},
-		}
-		i++
-	}
-
-	for _, j := range jobs {
-		if runArgs.once {
-			j.RepeatStrategy = jobrun.NoRepeatStrategy{}
-		}
-		if runArgs.job != "" {
-			if runArgs.job == j.Name {
-				runner.AddJob(j)
-				break
-			}
-			continue
-		}
-		runner.AddJob(j)
-	}
-
-	for {
-		select {
-		case job := <-runner.NotificationChan():
-			log.Printf("job %s reported error: %v\n", job.Name, job.LastError)
-		}
-	}
-
-	wg.Wait()
-
-}
-
-type localPullACL struct{}
-
-func (a localPullACL) Filter(p *zfs.DatasetPath) (pass bool, err error) {
-	return true, nil
-}
-
-func jobPull(pull *Pull, log jobrun.Logger) (err error) {
-
-	if lt, ok := pull.From.Transport.(LocalTransport); ok {
-
-		lt.SetHandler(Handler{
-			Logger: log,
-			// Allow access to any dataset since we control what mapping
-			// is passed to the pull routine.
-			// All local datasets will be passed to its Map() function,
-			// but only those for which a mapping exists will actually be pulled.
-			// We can pay this small performance penalty for now.
-			PullACL: localPullACL{},
-		})
-		pull.From.Transport = lt
-		log.Printf("fixing up local transport: %#v", pull.From.Transport)
-	}
-
-	var remote rpc.RPCRequester
-
-	if remote, err = pull.From.Transport.Connect(log); err != nil {
-		return
-	}
-
-	defer closeRPCWithTimeout(log, remote, time.Second*10, "")
-
-	return doPull(PullContext{remote, log, pull.Mapping, pull.InitialReplPolicy})
-}
-
-func jobPush(push *Push, log jobrun.Logger) (err error) {
-
-	if _, ok := push.To.Transport.(LocalTransport); ok {
-		panic("no support for local pushs")
-	}
-
-	var remote rpc.RPCRequester
-	if remote, err = push.To.Transport.Connect(log); err != nil {
-		return err
-	}
-
-	defer closeRPCWithTimeout(log, remote, time.Second*10, "")
-
-	log.Printf("building handler for PullMeRequest")
-	handler := Handler{
-		Logger:          log,
-		PullACL:         push.Filter,
-		SinkMappingFunc: nil, // no need for that in the handler for PullMe
-	}
-	log.Printf("handler: %#v", handler)
-
-	r := rpc.PullMeRequest{
-		InitialReplPolicy: push.InitialReplPolicy,
-	}
-	log.Printf("doing PullMeRequest: %#v", r)
-
-	if err = remote.PullMeRequest(r, handler); err != nil {
-		log.Printf("PullMeRequest failed: %s", err)
-		return
-	}
-
-	log.Printf("push job finished")
-	return
-
-}
-
-func closeRPCWithTimeout(log Logger, remote rpc.RPCRequester, timeout time.Duration, goodbye string) {
-	log.Printf("closing rpc connection")
-
-	ch := make(chan error)
-	go func() {
-		ch <- remote.CloseRequest(rpc.CloseRequest{goodbye})
-		close(ch)
-	}()
-
-	var err error
-	select {
-	case <-time.After(timeout):
-		err = fmt.Errorf("timeout exceeded (%s)", timeout)
-	case closeRequestErr := <-ch:
-		err = closeRequestErr
-	}
-
-	if err != nil {
-		log.Printf("error closing connection: %s", err)
-		err = remote.ForceClose()
-		if err != nil {
-			log.Printf("error force-closing connection: %s", err)
-		}
-	}
-	return
-}
-
-type PullContext struct {
-	Remote            rpc.RPCRequester
-	Log               Logger
-	Mapping           DatasetMapping
-	InitialReplPolicy rpc.InitialReplPolicy
-}
-
-func doPull(pull PullContext) (err error) {
-
-	remote := pull.Remote
-	log := pull.Log
-
-	log.Printf("requesting remote filesystem list")
-	fsr := rpc.FilesystemRequest{}
-	var remoteFilesystems []*zfs.DatasetPath
-	if remoteFilesystems, err = remote.FilesystemRequest(fsr); err != nil {
-		return
-	}
-
-	log.Printf("map remote filesystems to local paths and determine order for per-filesystem sync")
-	type RemoteLocalMapping struct {
-		Remote *zfs.DatasetPath
-		Local  *zfs.DatasetPath
-	}
-	replMapping := make(map[string]RemoteLocalMapping, len(remoteFilesystems))
-	localTraversal := zfs.NewDatasetPathForest()
-	for fs := range remoteFilesystems {
-		var err error
-		var localFs *zfs.DatasetPath
-		localFs, err = pull.Mapping.Map(remoteFilesystems[fs])
-		if err != nil {
-			if err != NoMatchError {
-				err := fmt.Errorf("error mapping %s: %s", remoteFilesystems[fs], err)
-				log.Printf("%s", err)
-				return err
-			}
-			continue
-		}
-		log.Printf("%s => %s", remoteFilesystems[fs].ToString(), localFs.ToString())
-		m := RemoteLocalMapping{remoteFilesystems[fs], localFs}
-		replMapping[m.Local.ToString()] = m
-		localTraversal.Add(m.Local)
-	}
-
-	log.Printf("build cache for already present local filesystem state")
-	localFilesystemState, err := zfs.ZFSListFilesystemState()
-	if err != nil {
-		log.Printf("error requesting local filesystem state: %s", err)
-		return err
-	}
-
-	log.Printf("start per-filesystem sync")
-	localTraversal.WalkTopDown(func(v zfs.DatasetPathVisit) bool {
-
-		if v.FilledIn {
-			if _, exists := localFilesystemState[v.Path.ToString()]; exists {
-				// No need to verify if this is a placeholder or not. It is sufficient
-				// to know we can add child filesystems to it
-				return true
-			}
-			log.Printf("creating placeholder filesystem %s", v.Path.ToString())
-			err = zfs.ZFSCreatePlaceholderFilesystem(v.Path)
-			if err != nil {
-				err = fmt.Errorf("aborting, cannot create placeholder filesystem %s: %s", v.Path, err)
-				return false
-			}
-			return true
-		}
-
-		m, ok := replMapping[v.Path.ToString()]
-		if !ok {
-			panic("internal inconsistency: replMapping should contain mapping for any path that was not filled in by WalkTopDown()")
-		}
-
-		log := func(format string, args ...interface{}) {
-			log.Printf("[%s => %s]: %s", m.Remote.ToString(), m.Local.ToString(), fmt.Sprintf(format, args...))
-		}
-
-		log("examing local filesystem state")
-		localState, localExists := localFilesystemState[m.Local.ToString()]
-		var versions []zfs.FilesystemVersion
-		switch {
-		case !localExists:
-			log("local filesystem does not exist")
-		case localState.Placeholder:
-			log("local filesystem is marked as placeholder")
-		default:
-			log("local filesystem exists")
-			log("requesting local filesystem versions")
-			if versions, err = zfs.ZFSListFilesystemVersions(m.Local, nil); err != nil {
-				log("cannot get local filesystem versions: %s", err)
-				return false
-			}
-		}
-
-		log("requesting remote filesystem versions")
-		var theirVersions []zfs.FilesystemVersion
-		theirVersions, err = remote.FilesystemVersionsRequest(rpc.FilesystemVersionsRequest{
-			Filesystem: m.Remote,
-		})
-		if err != nil {
-			log("error requesting remote filesystem versions: %s", err)
-			log("stopping replication for all filesystems mapped as children of %s", m.Local.ToString())
-			return false
-		}
-
-		log("computing diff between remote and local filesystem versions")
-		diff := zfs.MakeFilesystemDiff(versions, theirVersions)
-		log("%s", diff)
-
-		if localState.Placeholder && diff.Conflict != zfs.ConflictAllRight {
-			panic("internal inconsistency: local placeholder implies ConflictAllRight")
-		}
-
-		switch diff.Conflict {
-		case zfs.ConflictAllRight:
-
-			log("performing initial sync, following policy: '%s'", pull.InitialReplPolicy)
-
-			if pull.InitialReplPolicy != rpc.InitialReplPolicyMostRecent {
-				panic(fmt.Sprintf("policy '%s' not implemented", pull.InitialReplPolicy))
-			}
-
-			snapsOnly := make([]zfs.FilesystemVersion, 0, len(diff.MRCAPathRight))
-			for s := range diff.MRCAPathRight {
-				if diff.MRCAPathRight[s].Type == zfs.Snapshot {
-					snapsOnly = append(snapsOnly, diff.MRCAPathRight[s])
-				}
-			}
-
-			if len(snapsOnly) < 1 {
-				log("cannot perform initial sync: no remote snapshots. stopping...")
-				return false
-			}
-
-			r := rpc.InitialTransferRequest{
-				Filesystem:        m.Remote,
-				FilesystemVersion: snapsOnly[len(snapsOnly)-1],
-			}
-
-			log("requesting snapshot stream for %s", r.FilesystemVersion)
-
-			var stream io.Reader
-			if stream, err = remote.InitialTransferRequest(r); err != nil {
-				log("error requesting initial transfer: %s", err)
-				return false
-			}
-			log("received initial transfer request response")
-
-			log("invoking zfs receive")
-			watcher := util.IOProgressWatcher{Reader: stream}
-			watcher.KickOff(1*time.Second, func(p util.IOProgress) {
-				log("progress on receive operation: %v bytes received", p.TotalRX)
-			})
-
-			recvArgs := []string{"-u"}
-			if localState.Placeholder {
-				log("receive with forced rollback to replace placeholder filesystem")
-				recvArgs = append(recvArgs, "-F")
-			}
-
-			if err = zfs.ZFSRecv(m.Local, &watcher, recvArgs...); err != nil {
-				log("error receiving stream: %s", err)
-				return false
-			}
-			log("finished receiving stream, %v bytes total", watcher.Progress().TotalRX)
-
-			log("configuring properties of received filesystem")
-			if err = zfs.ZFSSet(m.Local, "readonly", "on"); err != nil {
-
-			}
-
-			log("finished initial transfer")
-			return true
-
-		case zfs.ConflictIncremental:
-
-			if len(diff.IncrementalPath) < 2 {
-				log("remote and local are in sync")
-				return true
-			}
-
-			log("following incremental path from diff")
-			var pathRx uint64
-
-			for i := 0; i < len(diff.IncrementalPath)-1; i++ {
-
-				from, to := diff.IncrementalPath[i], diff.IncrementalPath[i+1]
-
-				log := func(format string, args ...interface{}) {
-					log("[%v/%v][%s => %s]: %s", i+1, len(diff.IncrementalPath)-1,
-						from.Name, to.Name, fmt.Sprintf(format, args...))
-				}
-
-				log("requesting incremental snapshot stream")
-				r := rpc.IncrementalTransferRequest{
-					Filesystem: m.Remote,
-					From:       from,
-					To:         to,
-				}
-				var stream io.Reader
-				if stream, err = remote.IncrementalTransferRequest(r); err != nil {
-					log("error requesting incremental snapshot stream: %s", err)
-					return false
-				}
-
-				log("invoking zfs receive")
-				watcher := util.IOProgressWatcher{Reader: stream}
-				watcher.KickOff(1*time.Second, func(p util.IOProgress) {
-					log("progress on receive operation: %v bytes received", p.TotalRX)
-				})
-
-				if err = zfs.ZFSRecv(m.Local, &watcher); err != nil {
-					log("error receiving stream: %s", err)
-					return false
-				}
-
-				totalRx := watcher.Progress().TotalRX
-				pathRx += totalRx
-				log("finished incremental transfer, %v bytes total", totalRx)
-
-			}
-
-			log("finished following incremental path, %v bytes total", pathRx)
-			return true
-
-		case zfs.ConflictNoCommonAncestor:
-
-			log("remote and local filesystem have snapshots, but no common one")
-			log("perform manual replication to establish a common snapshot history")
-			log("remote versions:")
-			for _, v := range diff.MRCAPathRight {
-				log(" %s (GUID %v)", v, v.Guid)
-			}
-			log("local versions:")
-			for _, v := range diff.MRCAPathLeft {
-				log(" %s (GUID %v)", v, v.Guid)
-			}
-			return false
-
-		case zfs.ConflictDiverged:
-
-			log("remote and local filesystem share a history but have diverged")
-			log("perform manual replication or delete snapshots on the receiving" +
-				"side  to establish an incremental replication parse")
-			log("remote-only versions:")
-			for _, v := range diff.MRCAPathRight {
-				log(" %s (GUID %v)", v, v.Guid)
-			}
-			log("local-only versions:")
-			for _, v := range diff.MRCAPathLeft {
-				log(" %s (GUID %v)", v, v.Guid)
-			}
-			return false
-
-		}
-
-		panic("implementation error: this should not be reached")
-		return false
-
-	})
-
-	return
-
-}

cmd/sampleconf/zrepl.yml

@@ -1,109 +0,0 @@
-remotes:
-  offsite_backups:
-    transport:
-      ssh:
-        host: 192.168.122.6
-        user: root
-        port: 22
-        identity_file: /etc/zrepl/identities/offsite_backups
-
-pushs:
-
-  offsite:
-    to: offsite_backups
-    filter: {
-      # like in pull_acls
-      "tank/var/db<":   ok,
-      "tank/usr/home<": ok,
-    }
-
-pulls:
-
-  offsite:
-    from: offsite_backups
-    mapping: {
-      # like in sinks
-    }
-
-  # local replication, only allowed in pull mode
-  # the from name 'local' is reserved for this purpose
-  homemirror:
-    from: local
-    repeat:
-      interval: 15m
-    mapping: {
-      "tank/usr/home":"mirrorpool/foo/bar"
-    }
-
-sinks:
-
-  db1:
-    mapping: {
-      # direct mapping
-      "ssdpool/var/db/postgresql9.6":"zroot/backups/db1/pg_data"
-    }
-
-  mirror1:
-    mapping: {
-      # "<" subtree wildcard matches the dataset left of < and all its children
-      "tank/foo/bar<":"zroot/backups/mirror1" 
-    }
-
-  mirror2:
-    # more specific path patterns win over less specific ones
-    # direct mappings win over subtree wildcards
-    # detailed rule precedence: check unit tests & docs for exact behavior
-    # TODO subcommand to test a mapping & filter
-    mapping: {
-      "tank<":                       "zroot/backups/mirror1/tank1",
-      "tank/cdn/root<":              "storage/cdn/root",
-      "tank/legacydb":               "legacypool/backups/legacydb",
-    }
-
-
-pull_acls:
-
-  office_backup:
-    filter: {
-      # valid filter results (right hand side): ok, omit
-      # default is to omit
-      # rule precedence is same as for mappings
-      "tank<":         omit,
-      "tank/usr/home": ok,
-    }
-
-
-prune:
-
-  clean_backups:
-    policy: grid
-    grid: 6x10m | 24x1h | 7x1d | 5 x 1w | 4 x 5w
-    dataset_filter: {
-      "tank/backups/legacyscript<": omit,
-      "tank/backups<":              ok,
-    }
-    snapshot_filter: {
-      prefix: zrepl_
-    }
-
-  hfbak_prune: # cleans up after hfbak autosnap job
-    policy: grid
-    grid: 1x1m(keep=all)
-    dataset_filter: {
-      "pool1<": ok
-    }
-    snapshot_filter: {
-      prefix: zrepl_hfbak_
-    }
-
-autosnap:
-
-  hfbak:
-    prefix: zrepl_hfbak_
-    interval: 1s
-    dataset_filter: {
-      "pool1<": ok
-    }
-    # prune: hfbak_prune
-    # future versions may inline the retention policy here, but for now,
-    # pruning has to be triggered manually (it's safe to run autosnap + prune in parallel)

cmd/stdinserver.go

@@ -1,72 +0,0 @@
-package cmd
-
-import (
-	"fmt"
-	"github.com/spf13/cobra"
-	"github.com/zrepl/zrepl/rpc"
-	"github.com/zrepl/zrepl/sshbytestream"
-	"io"
-	golog "log"
-	"os"
-)
-
-var StdinserverCmd = &cobra.Command{
-	Use:   "stdinserver CLIENT_IDENTITY",
-	Short: "start in stdin server mode (from authorized_keys file)",
-	Run:   cmdStdinServer,
-}
-
-func init() {
-	RootCmd.AddCommand(StdinserverCmd)
-}
-
-func cmdStdinServer(cmd *cobra.Command, args []string) {
-
-	var err error
-	defer func() {
-		if err != nil {
-			log.Printf("stdinserver exiting with error: %s", err)
-			os.Exit(1)
-		}
-	}()
-
-	if len(args) != 1 || args[0] == "" {
-		err = fmt.Errorf("must specify client identity as positional argument")
-		return
-	}
-	identity := args[0]
-
-	pullACL, ok := conf.PullACLs[identity]
-	if !ok {
-		err = fmt.Errorf("could not find PullACL for identity '%s'", identity)
-		return
-	}
-
-	var sshByteStream io.ReadWriteCloser
-	if sshByteStream, err = sshbytestream.Incoming(); err != nil {
-		return
-	}
-
-	sinkMapping := func(identity string) (m DatasetMapping, err error) {
-		sink, ok := conf.Sinks[identity]
-		if !ok {
-			return nil, fmt.Errorf("could not find sink for identity '%s'", identity)
-		}
-		return sink, nil
-	}
-
-	sinkLogger := golog.New(logOut, fmt.Sprintf("sink[%s] ", identity), logFlags)
-	handler := Handler{
-		Logger:          sinkLogger,
-		SinkMappingFunc: sinkMapping,
-		PullACL:         pullACL,
-	}
-
-	if err = rpc.ListenByteStreamRPC(sshByteStream, identity, handler, sinkLogger); err != nil {
-		log.Printf("listenbytestreamerror: %#v\n", err)
-		os.Exit(1)
-	}
-
-	return
-
-}

config/config.go

@@ -0,0 +1,483 @@
+package config
+
+import (
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/zrepl/yaml-config"
+	"io/ioutil"
+	"os"
+	"reflect"
+	"regexp"
+	"strconv"
+	"time"
+)
+
+type Config struct {
+	Jobs   []JobEnum `yaml:"jobs"`
+	Global *Global   `yaml:"global,optional,fromdefaults"`
+}
+
+func (c *Config) Job(name string) (*JobEnum, error) {
+	for _, j := range c.Jobs {
+		if j.Name() == name {
+			return &j, nil
+		}
+	}
+	return nil, fmt.Errorf("job %q not defined in config", name)
+}
+
+type JobEnum struct {
+	Ret interface{}
+}
+
+func (j JobEnum) Name() string {
+	var name string
+	switch v := j.Ret.(type) {
+	case *PushJob: name = v.Name
+	case *SinkJob: name = v.Name
+	case *PullJob: name = v.Name
+	case *SourceJob: name = v.Name
+	default:
+		panic(fmt.Sprintf("unknownn job type %T", v))
+	}
+	return name
+}
+
+type ActiveJob struct {
+	Type         string                `yaml:"type"`
+	Name         string                `yaml:"name"`
+	Connect     ConnectEnum     `yaml:"connect"`
+	Pruning      PruningSenderReceiver `yaml:"pruning"`
+	Debug        JobDebugSettings      `yaml:"debug,optional"`
+}
+
+type PassiveJob struct {
+	Type        string           `yaml:"type"`
+	Name        string           `yaml:"name"`
+	Serve       ServeEnum `yaml:"serve"`
+	Debug       JobDebugSettings `yaml:"debug,optional"`
+}
+
+type PushJob struct {
+	ActiveJob `yaml:",inline"`
+	Snapshotting SnapshottingEnum          `yaml:"snapshotting"`
+	Filesystems FilesystemsFilter `yaml:"filesystems"`
+}
+
+type PullJob struct {
+	ActiveJob `yaml:",inline"`
+	RootFS    string        `yaml:"root_fs"`
+	Interval  time.Duration `yaml:"interval,positive"`
+}
+
+type SinkJob struct {
+	PassiveJob `yaml:",inline"`
+	RootFS     string `yaml:"root_fs"`
+}
+
+type SourceJob struct {
+	PassiveJob `yaml:",inline"`
+	Snapshotting SnapshottingEnum      `yaml:"snapshotting"`
+	Filesystems FilesystemsFilter `yaml:"filesystems"`
+}
+
+type FilesystemsFilter map[string]bool
+
+type SnapshottingEnum struct {
+	Ret interface{}
+}
+
+type SnapshottingPeriodic struct {
+	Type string		`yaml:"type"`
+	Prefix string	`yaml:"prefix"`
+	Interval time.Duration `yaml:"interval,positive"`
+}
+
+type SnapshottingManual struct {
+	Type string `yaml:"type"`
+}
+
+type PruningSenderReceiver struct {
+	KeepSender   []PruningEnum `yaml:"keep_sender"`
+	KeepReceiver []PruningEnum `yaml:"keep_receiver"`
+}
+
+type PruningLocal struct {
+	Keep []PruningEnum `yaml:"keep"`
+}
+
+type LoggingOutletEnumList []LoggingOutletEnum
+
+func (l *LoggingOutletEnumList) SetDefault() {
+	def := `
+type: "stdout"
+time: true
+level: "warn"
+format: "human"
+`
+	s := &StdoutLoggingOutlet{}
+	err := yaml.UnmarshalStrict([]byte(def), &s)
+	if err != nil {
+		panic(err)
+	}
+	*l = []LoggingOutletEnum{LoggingOutletEnum{Ret: s}}
+}
+
+var _ yaml.Defaulter = &LoggingOutletEnumList{}
+
+type Global struct {
+	Logging    *LoggingOutletEnumList `yaml:"logging,optional,fromdefaults"`
+	Monitoring []MonitoringEnum       `yaml:"monitoring,optional"`
+	Control    *GlobalControl         `yaml:"control,optional,fromdefaults"`
+	Serve      *GlobalServe           `yaml:"serve,optional,fromdefaults"`
+	RPC        *RPCConfig             `yaml:"rpc,optional,fromdefaults"`
+}
+
+func Default(i interface{}) {
+	v := reflect.ValueOf(i)
+	if v.Kind() != reflect.Ptr {
+		panic(v)
+	}
+	y := `{}`
+	err := yaml.Unmarshal([]byte(y), v.Interface())
+	if err != nil {
+		panic(err)
+	}
+}
+
+type RPCConfig struct {
+	Timeout             time.Duration `yaml:"timeout,optional,positive,default=10s"`
+	TxChunkSize         uint32        `yaml:"tx_chunk_size,optional,default=32768"`
+	RxStructuredMaxLen  uint32        `yaml:"rx_structured_max,optional,default=16777216"`
+	RxStreamChunkMaxLen uint32        `yaml:"rx_stream_chunk_max,optional,default=16777216"`
+	RxHeaderMaxLen      uint32        `yaml:"rx_header_max,optional,default=40960"`
+	SendHeartbeatInterval             time.Duration `yaml:"send_heartbeat_interval,optional,positive,default=5s"`
+
+}
+
+type ConnectEnum struct {
+	Ret interface{}
+}
+
+type ConnectCommon struct {
+	Type string     `yaml:"type"`
+	RPC  *RPCConfig `yaml:"rpc,optional"`
+}
+
+type TCPConnect struct {
+	ConnectCommon `yaml:",inline"`
+	Address       string        `yaml:"address"`
+	DialTimeout   time.Duration `yaml:"dial_timeout,positive,default=10s"`
+}
+
+type TLSConnect struct {
+	ConnectCommon `yaml:",inline"`
+	Address       string        `yaml:"address"`
+	Ca            string        `yaml:"ca"`
+	Cert          string        `yaml:"cert"`
+	Key           string        `yaml:"key"`
+	ServerCN      string        `yaml:"server_cn"`
+	DialTimeout   time.Duration `yaml:"dial_timeout,positive,default=10s"`
+}
+
+type SSHStdinserverConnect struct {
+	ConnectCommon        `yaml:",inline"`
+	Host                 string        `yaml:"host"`
+	User                 string        `yaml:"user"`
+	Port                 uint16        `yaml:"port"`
+	IdentityFile         string        `yaml:"identity_file"`
+	TransportOpenCommand []string      `yaml:"transport_open_command,optional"` //TODO unused
+	SSHCommand           string        `yaml:"ssh_command,optional"`            //TODO unused
+	Options              []string      `yaml:"options,optional"`
+	DialTimeout          time.Duration `yaml:"dial_timeout,positive,default=10s"`
+}
+
+type LocalConnect struct {
+	ConnectCommon `yaml:",inline"`
+	ListenerName string `yaml:"listener_name"`
+	ClientIdentity string `yaml:"client_identity"`
+}
+
+type ServeEnum struct {
+	Ret interface{}
+}
+
+type ServeCommon struct {
+	Type string     `yaml:"type"`
+	RPC  *RPCConfig `yaml:"rpc,optional"`
+}
+
+type TCPServe struct {
+	ServeCommon `yaml:",inline"`
+	Listen      string            `yaml:"listen"`
+	Clients     map[string]string `yaml:"clients"`
+}
+
+type TLSServe struct {
+	ServeCommon      `yaml:",inline"`
+	Listen           string        `yaml:"listen"`
+	Ca               string        `yaml:"ca"`
+	Cert             string        `yaml:"cert"`
+	Key              string        `yaml:"key"`
+	ClientCNs        []string      `yaml:"client_cns"`
+	HandshakeTimeout time.Duration `yaml:"handshake_timeout,positive,default=10s"`
+}
+
+type StdinserverServer struct {
+	ServeCommon    `yaml:",inline"`
+	ClientIdentities []string `yaml:"client_identities"`
+}
+
+type LocalServe struct {
+	ServeCommon `yaml:",inline"`
+	ListenerName string `yaml:"listener_name"`
+}
+
+type PruningEnum struct {
+	Ret interface{}
+}
+
+type PruneKeepNotReplicated struct {
+	Type string `yaml:"type"`
+	KeepSnapshotAtCursor bool `yaml:"keep_snapshot_at_cursor,optional,default=true"`
+}
+
+type PruneKeepLastN struct {
+	Type  string `yaml:"type"`
+	Count int    `yaml:"count"`
+}
+
+type PruneKeepRegex struct { // FIXME rename to KeepRegex
+	Type  string `yaml:"type"`
+	Regex string `yaml:"regex"`
+}
+
+type LoggingOutletEnum struct {
+	Ret interface{}
+}
+
+type LoggingOutletCommon struct {
+	Type   string `yaml:"type"`
+	Level  string `yaml:"level"`
+	Format string `yaml:"format"`
+}
+
+type StdoutLoggingOutlet struct {
+	LoggingOutletCommon `yaml:",inline"`
+	Time                bool `yaml:"time,default=true"`
+	Color               bool `yaml:"color,default=true"`
+}
+
+type SyslogLoggingOutlet struct {
+	LoggingOutletCommon `yaml:",inline"`
+	RetryInterval       time.Duration `yaml:"retry_interval,positive,default=10s"`
+}
+
+type TCPLoggingOutlet struct {
+	LoggingOutletCommon `yaml:",inline"`
+	Address             string               `yaml:"address"`
+	Net                 string               `yaml:"net,default=tcp"`
+	RetryInterval       time.Duration        `yaml:"retry_interval,positive,default=10s"`
+	TLS                 *TCPLoggingOutletTLS `yaml:"tls,optional"`
+}
+
+type TCPLoggingOutletTLS struct {
+	CA   string `yaml:"ca"`
+	Cert string `yaml:"cert"`
+	Key  string `yaml:"key"`
+}
+
+type MonitoringEnum struct {
+	Ret interface{}
+}
+
+type PrometheusMonitoring struct {
+	Type   string `yaml:"type"`
+	Listen string `yaml:"listen"`
+}
+
+type GlobalControl struct {
+	SockPath string `yaml:"sockpath,default=/var/run/zrepl/control"`
+}
+
+type GlobalServe struct {
+	StdinServer *GlobalStdinServer `yaml:"stdinserver,optional,fromdefaults"`
+}
+
+type GlobalStdinServer struct {
+	SockDir string `yaml:"sockdir,default=/var/run/zrepl/stdinserver"`
+}
+
+type JobDebugSettings struct {
+	Conn *struct {
+		ReadDump  string `yaml:"read_dump"`
+		WriteDump string `yaml:"write_dump"`
+	} `yaml:"conn,optional"`
+	RPCLog bool `yaml:"rpc_log,optional,default=false"`
+}
+
+func enumUnmarshal(u func(interface{}, bool) error, types map[string]interface{}) (interface{}, error) {
+	var in struct {
+		Type string
+	}
+	if err := u(&in, true); err != nil {
+		return nil, err
+	}
+	if in.Type == "" {
+		return nil, &yaml.TypeError{Errors: []string{"must specify type"}}
+	}
+
+	v, ok := types[in.Type]
+	if !ok {
+		return nil, &yaml.TypeError{Errors: []string{fmt.Sprintf("invalid type name %q", in.Type)}}
+	}
+	if err := u(v, false); err != nil {
+		return nil, err
+	}
+	return v, nil
+}
+
+func (t *JobEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"push":   &PushJob{},
+		"sink":   &SinkJob{},
+		"pull":   &PullJob{},
+		"source": &SourceJob{},
+	})
+	return
+}
+
+func (t *ConnectEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"tcp":             &TCPConnect{},
+		"tls":             &TLSConnect{},
+		"ssh+stdinserver": &SSHStdinserverConnect{},
+		"local": 		   &LocalConnect{},
+	})
+	return
+}
+
+func (t *ServeEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"tcp":         &TCPServe{},
+		"tls":         &TLSServe{},
+		"stdinserver": &StdinserverServer{},
+		"local"      : &LocalServe{},
+	})
+	return
+}
+
+func (t *PruningEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"not_replicated": &PruneKeepNotReplicated{},
+		"last_n":         &PruneKeepLastN{},
+		"grid":           &PruneGrid{},
+		"regex":          &PruneKeepRegex{},
+	})
+	return
+}
+
+func (t *SnapshottingEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"periodic": &SnapshottingPeriodic{},
+		"manual": &SnapshottingManual{},
+	})
+	return
+}
+
+func (t *LoggingOutletEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"stdout": &StdoutLoggingOutlet{},
+		"syslog": &SyslogLoggingOutlet{},
+		"tcp":    &TCPLoggingOutlet{},
+	})
+	return
+}
+
+func (t *MonitoringEnum) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	t.Ret, err = enumUnmarshal(u, map[string]interface{}{
+		"prometheus": &PrometheusMonitoring{},
+	})
+	return
+}
+
+var ConfigFileDefaultLocations = []string{
+	"/etc/zrepl/zrepl.yml",
+	"/usr/local/etc/zrepl/zrepl.yml",
+}
+
+func ParseConfig(path string) (i *Config, err error) {
+
+	if path == "" {
+		// Try default locations
+		for _, l := range ConfigFileDefaultLocations {
+			stat, statErr := os.Stat(l)
+			if statErr != nil {
+				continue
+			}
+			if !stat.Mode().IsRegular() {
+				err = errors.Errorf("file at default location is not a regular file: %s", l)
+				return
+			}
+			path = l
+			break
+		}
+	}
+
+	var bytes []byte
+
+	if bytes, err = ioutil.ReadFile(path); err != nil {
+		return
+	}
+
+	return ParseConfigBytes(bytes)
+}
+
+func ParseConfigBytes(bytes []byte) (*Config, error) {
+	var c *Config
+	if err := yaml.UnmarshalStrict(bytes, &c); err != nil {
+		return nil, err
+	}
+	if c == nil {
+		return nil, fmt.Errorf("config is empty or only consists of comments")
+	}
+	return c, nil
+}
+
+var durationStringRegex *regexp.Regexp = regexp.MustCompile(`^\s*(\d+)\s*(s|m|h|d|w)\s*$`)
+
+func parsePostitiveDuration(e string) (d time.Duration, err error) {
+	comps := durationStringRegex.FindStringSubmatch(e)
+	if len(comps) != 3 {
+		err = fmt.Errorf("does not match regex: %s %#v", e, comps)
+		return
+	}
+
+	durationFactor, err := strconv.ParseInt(comps[1], 10, 64)
+	if err != nil {
+		return 0, err
+	}
+	if durationFactor <= 0 {
+		return 0, errors.New("duration must be positive integer")
+	}
+
+	var durationUnit time.Duration
+	switch comps[2] {
+	case "s":
+		durationUnit = time.Second
+	case "m":
+		durationUnit = time.Minute
+	case "h":
+		durationUnit = time.Hour
+	case "d":
+		durationUnit = 24 * time.Hour
+	case "w":
+		durationUnit = 24 * 7 * time.Hour
+	default:
+		err = fmt.Errorf("contains unknown time unit '%s'", comps[2])
+		return
+	}
+
+	d = time.Duration(durationFactor) * durationUnit
+	return
+}

config/config_global_test.go

@@ -0,0 +1,82 @@
+package config
+
+import (
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/zrepl/yaml-config"
+	"testing"
+)
+
+func testValidGlobalSection(t *testing.T, s string) *Config {
+	jobdef := `
+jobs:
+- name: dummyjob
+  type: sink
+  serve:
+    type: tcp
+    listen: ":2342"
+    clients: {
+      "10.0.0.1":"foo"
+    }
+  root_fs: zoot/foo
+`
+	_, err := ParseConfigBytes([]byte(jobdef))
+	require.NoError(t, err)
+	return testValidConfig(t, s + jobdef)
+}
+
+func TestOutletTypes(t *testing.T) {
+	conf := testValidGlobalSection(t, `
+global:
+  logging:
+  - type: stdout
+    level: debug
+    format: human
+  - type: syslog
+    level: info
+    retry_interval: 20s
+    format: human
+  - type: tcp
+    level: debug
+    format: json
+    address: logserver.example.com:1234
+  - type: tcp
+    level: debug
+    format: json
+    address: encryptedlogserver.example.com:1234
+    retry_interval: 20s 
+    tls:
+      ca: /etc/zrepl/log/ca.crt
+      cert: /etc/zrepl/log/key.pem
+      key: /etc/zrepl/log/cert.pem
+`)
+	assert.Equal(t, 4, len(*conf.Global.Logging))
+	assert.NotNil(t, (*conf.Global.Logging)[3].Ret.(*TCPLoggingOutlet).TLS)
+}
+
+func TestDefaultLoggingOutlet(t *testing.T) {
+	conf := testValidGlobalSection(t, "")
+	assert.Equal(t, 1, len(*conf.Global.Logging))
+	o := (*conf.Global.Logging)[0].Ret.(*StdoutLoggingOutlet)
+	assert.Equal(t, "warn", o.Level)
+	assert.Equal(t, "human", o.Format)
+}
+
+func TestPrometheusMonitoring(t *testing.T) {
+	conf := testValidGlobalSection(t, `
+global:
+  monitoring:
+    - type: prometheus
+      listen: ':9091'
+`)
+	assert.Equal(t, ":9091", conf.Global.Monitoring[0].Ret.(*PrometheusMonitoring).Listen)	
+}
+
+func TestLoggingOutletEnumList_SetDefaults(t *testing.T) {
+	e := &LoggingOutletEnumList{}
+	var i yaml.Defaulter = e
+	require.NotPanics(t, func() {
+		i.SetDefault()
+		assert.Equal(t, "warn", (*e)[0].Ret.(*StdoutLoggingOutlet).Level)
+	})
+}

config/config_minimal_test.go

@@ -0,0 +1,39 @@
+package config
+
+import (
+	"testing"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestConfigEmptyFails(t *testing.T) {
+	conf, err := testConfig(t, "\n")
+	assert.Nil(t, conf)
+	assert.Error(t, err)
+}
+
+func TestJobsOnlyWorks(t *testing.T) {
+	testValidConfig(t, `
+jobs:
+- name: push
+  type: push
+  # snapshot the filesystems matched by the left-hand-side of the mapping
+  # every 10m with zrepl_ as prefix
+  connect:
+    type: tcp
+    address: localhost:2342
+  filesystems: {
+    "pool1/var/db<": true,
+    "pool1/usr/home<": true,
+    "pool1/usr/home/paranoid": false, #don't backup paranoid user
+    "pool1/poudriere/ports<": false #don't backup the ports trees
+  }
+  snapshotting:
+    type: manual
+  pruning:
+    keep_sender:
+    - type: not_replicated
+    keep_receiver:
+    - type: last_n 
+      count: 1
+`)
+}

config/config_rpc_test.go

@@ -0,0 +1,86 @@
+package config
+
+import (
+	"github.com/stretchr/testify/assert"
+	"testing"
+	"time"
+)
+
+func TestRPC(t *testing.T) {
+	conf := testValidConfig(t, `
+jobs:
+- name: pull_servers
+  type: pull
+  connect:
+    type: tcp
+    address: "server1.foo.bar:8888"
+    rpc:
+      timeout: 20s # different form default, should merge
+  root_fs: "pool2/backup_servers"
+  interval: 10m
+  pruning:
+    keep_sender:
+      - type: not_replicated
+    keep_receiver:
+      - type: last_n
+        count: 100
+
+- name: pull_servers2
+  type: pull
+  connect:
+    type: tcp
+    address: "server1.foo.bar:8888"
+    rpc:
+      tx_chunk_size: 0xabcd # different from default, should merge
+  root_fs: "pool2/backup_servers"
+  interval: 10m
+  pruning:
+    keep_sender:
+    - type: not_replicated
+    keep_receiver:
+    - type: last_n
+      count: 100
+
+- type: sink
+  name: "laptop_sink"
+  root_fs: "pool2/backup_laptops"
+  serve:
+    type: tcp
+    listen: "192.168.122.189:8888"
+    clients: {
+	"10.23.42.23":"client1"
+    }
+    rpc:
+      rx_structured_max: 0x2342
+
+- type: sink
+  name: "other_sink"
+  root_fs: "pool2/backup_laptops"
+  serve:
+    type: tcp
+    listen: "192.168.122.189:8888"
+    clients: {
+	"10.23.42.23":"client1"
+    }
+    rpc:
+      send_heartbeat_interval: 10s
+
+`)
+
+	assert.Equal(t, 20*time.Second, conf.Jobs[0].Ret.(*PullJob).Connect.Ret.(*TCPConnect).RPC.Timeout)
+	assert.Equal(t, uint32(0xabcd), conf.Jobs[1].Ret.(*PullJob).Connect.Ret.(*TCPConnect).RPC.TxChunkSize)
+	assert.Equal(t, uint32(0x2342), conf.Jobs[2].Ret.(*SinkJob).Serve.Ret.(*TCPServe).RPC.RxStructuredMaxLen)
+	assert.Equal(t, 10*time.Second, conf.Jobs[3].Ret.(*SinkJob).Serve.Ret.(*TCPServe).RPC.SendHeartbeatInterval)
+	defConf := RPCConfig{}
+	Default(&defConf)
+	assert.Equal(t, defConf.Timeout, conf.Global.RPC.Timeout)
+}
+
+func TestGlobal_DefaultRPCConfig(t *testing.T) {
+	assert.NotPanics(t, func() {
+		var c RPCConfig
+		Default(&c)
+		assert.NotNil(t, c)
+		assert.Equal(t, c.TxChunkSize, uint32(1)<<15)
+	})
+}

config/config_snapshotting_test.go

@@ -0,0 +1,57 @@
+package config
+
+import (
+	"fmt"
+	"github.com/stretchr/testify/assert"
+	"testing"
+	"time"
+)
+
+func TestSnapshotting(t *testing.T) {
+	tmpl := `
+jobs:
+- name: foo
+  type: push
+  connect:
+    type: local
+    listener_name: foo
+    client_identity: bar
+  filesystems: {"<": true}
+  %s
+  pruning:
+    keep_sender:
+    - type: last_n
+      count: 10
+    keep_receiver:
+    - type: last_n
+      count: 10
+`
+	manual := `
+  snapshotting:
+    type: manual
+`
+	periodic := `
+  snapshotting:
+    type: periodic
+    prefix: zrepl_
+    interval: 10m
+`
+
+	fillSnapshotting := func(s string) string {return fmt.Sprintf(tmpl, s)}
+	var c *Config
+
+	t.Run("manual", func(t *testing.T) {
+		c = testValidConfig(t, fillSnapshotting(manual))
+		snm := c.Jobs[0].Ret.(*PushJob).Snapshotting.Ret.(*SnapshottingManual)
+		assert.Equal(t, "manual", snm.Type)
+	})
+
+	t.Run("periodic", func(t *testing.T) {
+		c = testValidConfig(t, fillSnapshotting(periodic))
+		snp := c.Jobs[0].Ret.(*PushJob).Snapshotting.Ret.(*SnapshottingPeriodic)
+		assert.Equal(t, "periodic", snp.Type)
+		assert.Equal(t, 10*time.Minute, snp.Interval)
+		assert.Equal(t, "zrepl_" , snp.Prefix)
+	})
+
+}

config/config_test.go

@@ -0,0 +1,50 @@
+package config
+
+import (
+	"github.com/kr/pretty"
+	"github.com/stretchr/testify/require"
+	"path"
+	"path/filepath"
+	"testing"
+)
+
+func TestSampleConfigsAreParsedWithoutErrors(t *testing.T) {
+	paths, err := filepath.Glob("./samples/*")
+	if err != nil {
+		t.Errorf("glob failed: %+v", err)
+	}
+
+	for _, p := range paths {
+
+		if path.Ext(p) != ".yml" {
+			t.Logf("skipping file %s", p)
+			continue
+		}
+
+		t.Run(p, func(t *testing.T) {
+			c, err := ParseConfig(p)
+			if err != nil {
+				t.Errorf("error parsing %s:\n%+v", p, err)
+			}
+
+			t.Logf("file: %s", p)
+			t.Log(pretty.Sprint(c))
+		})
+
+	}
+
+}
+
+
+func testValidConfig(t *testing.T, input string) (*Config) {
+	t.Helper()
+	conf, err := testConfig(t, input)
+	require.NoError(t, err)
+	require.NotNil(t, conf)
+	return conf
+}
+
+func testConfig(t *testing.T, input string) (*Config, error) {
+	t.Helper()
+	return ParseConfigBytes([]byte(input))
+}

config/retentiongrid.go

@@ -0,0 +1,123 @@
+package config
+
+import (
+	"fmt"
+	"regexp"
+	"strconv"
+	"strings"
+	"time"
+)
+
+type RetentionIntervalList []RetentionInterval
+
+type PruneGrid struct {
+	Type          string                `yaml:"type"`
+	Grid          RetentionIntervalList `yaml:"grid"`
+	Regex	      string                `yaml:"regex"`
+}
+
+type RetentionInterval struct {
+	length    time.Duration
+	keepCount int
+}
+
+func (i *RetentionInterval) Length() time.Duration {
+	return i.length
+}
+
+func (i *RetentionInterval) KeepCount() int {
+	return i.keepCount
+}
+
+const RetentionGridKeepCountAll int = -1
+
+type RetentionGrid struct {
+	intervals []RetentionInterval
+}
+
+func (t *RetentionIntervalList) UnmarshalYAML(u func(interface{}, bool) error) (err error) {
+	var in string
+	if err := u(&in, true); err != nil {
+		return err
+	}
+
+	intervals, err := parseRetentionGridIntervalsString(in)
+	if err != nil {
+		return err
+	}
+
+	*t = intervals
+
+	return nil
+}
+
+var retentionStringIntervalRegex *regexp.Regexp = regexp.MustCompile(`^\s*(\d+)\s*x\s*([^\(]+)\s*(\((.*)\))?\s*$`)
+
+func parseRetentionGridIntervalString(e string) (intervals []RetentionInterval, err error) {
+
+	comps := retentionStringIntervalRegex.FindStringSubmatch(e)
+	if comps == nil {
+		err = fmt.Errorf("retention string does not match expected format")
+		return
+	}
+
+	times, err := strconv.Atoi(comps[1])
+	if err != nil {
+		return nil, err
+	} else if times <= 0 {
+		return nil, fmt.Errorf("contains factor <= 0")
+	}
+
+	duration, err := parsePostitiveDuration(comps[2])
+	if err != nil {
+		return nil, err
+	}
+
+	keepCount := 1
+	if comps[3] != "" {
+		// Decompose key=value, comma separated
+		// For now, only keep_count is supported
+		re := regexp.MustCompile(`^\s*keep=(.+)\s*$`)
+		res := re.FindStringSubmatch(comps[4])
+		if res == nil || len(res) != 2 {
+			err = fmt.Errorf("interval parameter contains unknown parameters")
+			return
+		}
+		if res[1] == "all" {
+			keepCount = RetentionGridKeepCountAll
+		} else {
+			keepCount, err = strconv.Atoi(res[1])
+			if err != nil {
+				err = fmt.Errorf("cannot parse keep_count value")
+				return
+			}
+		}
+	}
+
+	intervals = make([]RetentionInterval, times)
+	for i := range intervals {
+		intervals[i] = RetentionInterval{
+			length:    duration,
+			keepCount: keepCount,
+		}
+	}
+
+	return
+
+}
+
+func parseRetentionGridIntervalsString(s string) (intervals []RetentionInterval, err error) {
+
+	ges := strings.Split(s, "|")
+	intervals = make([]RetentionInterval, 0, 7*len(ges))
+
+	for intervalIdx, e := range ges {
+		parsed, err := parseRetentionGridIntervalString(e)
+		if err != nil {
+			return nil, fmt.Errorf("cannot parse interval %d of %d: %s: %s", intervalIdx+1, len(ges), err, strings.TrimSpace(e))
+		}
+		intervals = append(intervals, parsed...)
+	}
+
+	return
+}

config/samples/local.yml

@@ -0,0 +1,31 @@
+
+jobs:
+  - type: sink
+    name: "local_sink"
+    root_fs: "storage/zrepl/sink"
+    serve:
+      type: local
+      listener_name: localsink
+
+  - type: push
+    name: "backup_system"
+    connect:
+      type: local
+      listener_name: localsink
+      client_identity: local_backup
+    filesystems: {
+      "system<": true,
+    }
+    snapshotting:
+      type: periodic
+      interval: 10m
+      prefix: zrepl_
+    pruning:
+      keep_sender:
+      - type: not_replicated
+      - type: last_n
+        count: 10
+      keep_receiver:
+      - type: grid
+        grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+        regex: "zrepl_.*"

config/samples/pull.yml

@@ -0,0 +1,24 @@
+jobs:
+- name: pull_servers
+  type: pull
+  connect:
+    type: tls
+    address: "server1.foo.bar:8888"
+    ca: "/certs/ca.crt"
+    cert: "/certs/cert.crt"
+    key: "/certs/key.pem"
+    server_cn: "server1"
+  root_fs: "pool2/backup_servers"
+  interval: 10m
+  pruning:
+    keep_sender:
+    - type: not_replicated
+    - type: last_n
+      count: 10
+    - type: grid
+      grid: 1x1h(keep=all) | 24x1h | 14x1d
+      regex: "zrepl_.*"
+    keep_receiver:
+    - type: grid
+      grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+      regex: "zrepl_.*"

config/samples/pull_ssh.yml

@@ -0,0 +1,28 @@
+jobs:
+
+- name: pull_servers
+  type: pull
+  connect:
+    type: ssh+stdinserver
+    host: app-srv.example.com
+    user: root
+    port: 22
+    identity_file: /etc/zrepl/ssh/identity
+    options: # optional, default [], `-o` arguments passed to ssh
+    - "Compression=on"
+  root_fs: "pool2/backup_servers"
+  interval: 10m
+  pruning:
+    keep_sender:
+    - type: not_replicated
+    - type: last_n
+      count: 10
+    - type: grid
+      grid: 1x1h(keep=all) | 24x1h | 14x1d
+      regex: "^zrepl_.*"
+    keep_receiver:
+    - type: regex
+      regex: keep_
+    - type: grid
+      grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+      regex: "^zrepl_.*"

config/samples/push.yml

@@ -0,0 +1,24 @@
+jobs:
+  - type: push
+    name: "push"
+    filesystems: {
+      "<": true,
+      "tmp": false
+    }
+    connect:
+      type: tcp
+      address: "backup-server.foo.bar:8888"
+    snapshotting:
+      type: manual
+    pruning:
+      keep_sender:
+        - type: not_replicated
+        - type: last_n
+          count: 10
+        - type: grid
+          grid: 1x1h(keep=all) | 24x1h | 14x1d
+          regex: "^zrepl_.*"
+      keep_receiver:
+        - type: grid
+          grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+          regex: "^zrepl_.*"

config/samples/sink.yml

@@ -0,0 +1,13 @@
+jobs:
+  - type: sink
+    name: "laptop_sink"
+    root_fs: "pool2/backup_laptops"
+    serve:
+      type: tls
+      listen: "192.168.122.189:8888"
+      ca: "ca.pem"
+      cert: "cert.pem"
+      key: "key.pem"
+      client_cns:
+        - "laptop1"
+        - "homeserver"

config/samples/source.yml

@@ -0,0 +1,17 @@
+jobs:
+- name: pull_source
+  type: source
+  serve:
+    type: tcp
+    listen: "0.0.0.0:8888"
+    clients: {
+      "192.168.122.123" : "client1"
+    }
+  filesystems: {
+    "<": true,
+    "secret": false
+  }
+  snapshotting:
+    type: periodic
+    interval: 10m
+    prefix: zrepl_

config/samples/source_ssh.yml

@@ -0,0 +1,17 @@
+jobs:
+- name: pull_source
+  type: source
+  serve:
+    type: stdinserver
+    client_identities:
+      - "client1"
+      - "client2"
+  filesystems: {
+    "<": true,
+    "secret": false
+  }
+  snapshotting:
+    type: periodic
+    interval: 10m
+    prefix: zrepl_
+

daemon/control.go

@@ -0,0 +1,241 @@
+package daemon
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/daemon/job"
+	"github.com/zrepl/zrepl/daemon/nethelpers"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/version"
+	"io"
+	"net"
+	"net/http"
+	"time"
+)
+
+type controlJob struct {
+	sockaddr *net.UnixAddr
+	jobs     *jobs
+}
+
+func newControlJob(sockpath string, jobs *jobs) (j *controlJob, err error) {
+	j = &controlJob{jobs: jobs}
+
+	j.sockaddr, err = net.ResolveUnixAddr("unix", sockpath)
+	if err != nil {
+		err = errors.Wrap(err, "cannot resolve unix address")
+		return
+	}
+
+	return
+}
+
+func (j *controlJob) Name() string { return jobNameControl }
+
+func (j *controlJob) Status() *job.Status { return &job.Status{Type: job.TypeInternal} }
+
+var promControl struct {
+	requestBegin *prometheus.CounterVec
+	requestFinished *prometheus.HistogramVec
+}
+
+func (j *controlJob) RegisterMetrics(registerer prometheus.Registerer) {
+	promControl.requestBegin = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Namespace:   "zrepl",
+		Subsystem:   "control",
+		Name:        "request_begin",
+		Help:        "number of request we started to handle",
+	}, []string{"endpoint"})
+
+	promControl.requestFinished = prometheus.NewHistogramVec(prometheus.HistogramOpts{
+		Namespace:   "zrepl",
+		Subsystem:   "control",
+		Name:        "request_finished",
+		Help:        "time it took a request to finih",
+		Buckets: []float64{1e-6, 10e-6, 100e-6, 500e-6, 1e-3,10e-3, 100e-3, 200e-3,400e-3,800e-3, 1, 10, 20},
+	}, []string{"endpoint"})
+	registerer.MustRegister(promControl.requestBegin)
+	registerer.MustRegister(promControl.requestFinished)
+}
+
+const (
+	ControlJobEndpointPProf   string = "/debug/pprof"
+	ControlJobEndpointVersion string = "/version"
+	ControlJobEndpointStatus  string = "/status"
+	ControlJobEndpointSignal  string = "/signal"
+)
+
+func (j *controlJob) Run(ctx context.Context) {
+
+	log := job.GetLogger(ctx)
+	defer log.Info("control job finished")
+
+	l, err := nethelpers.ListenUnixPrivate(j.sockaddr)
+	if err != nil {
+		log.WithError(err).Error("error listening")
+		return
+	}
+
+	pprofServer := NewPProfServer(ctx)
+
+	mux := http.NewServeMux()
+	mux.Handle(ControlJobEndpointPProf,
+		requestLogger{log: log, handler: jsonRequestResponder{func(decoder jsonDecoder) (interface{}, error) {
+			var msg PprofServerControlMsg
+			err := decoder(&msg)
+			if err != nil {
+				return nil, errors.Errorf("decode failed")
+			}
+			pprofServer.Control(msg)
+			return struct{}{}, nil
+		}}})
+
+	mux.Handle(ControlJobEndpointVersion,
+		requestLogger{log: log, handler: jsonResponder{func() (interface{}, error) {
+			return version.NewZreplVersionInformation(), nil
+		}}})
+
+	mux.Handle(ControlJobEndpointStatus,
+		// don't log requests to status endpoint, too spammy
+		jsonResponder{func() (interface{}, error) {
+			s := j.jobs.status()
+			return s, nil
+		}})
+
+	mux.Handle(ControlJobEndpointSignal,
+		requestLogger{log: log, handler: jsonRequestResponder{func(decoder jsonDecoder) (interface{}, error) {
+			type reqT struct {
+				Name string
+				Op string
+			}
+			var req reqT
+			if decoder(&req) != nil {
+				return nil, errors.Errorf("decode failed")
+			}
+
+			var err error
+			switch req.Op {
+			case "wakeup":
+				err = j.jobs.wakeup(req.Name)
+			case "reset":
+				err = j.jobs.reset(req.Name)
+			default:
+				err = fmt.Errorf("operation %q is invalid", req.Op)
+			}
+
+			return struct{}{}, err
+		}}})
+	server := http.Server{
+		Handler: mux,
+		// control socket is local, 1s timeout should be more than sufficient, even on a loaded system
+		WriteTimeout: 1*time.Second,
+		ReadTimeout: 1*time.Second,
+	}
+
+outer:
+	for {
+
+		served := make(chan error)
+		go func() {
+			served <- server.Serve(l)
+			close(served)
+		}()
+
+		select {
+		case <-ctx.Done():
+			log.WithError(ctx.Err()).Info("context done")
+			server.Shutdown(context.Background())
+			break outer
+		case err = <-served:
+			if err != nil {
+				log.WithError(err).Error("error serving")
+				break outer
+			}
+		}
+
+	}
+
+}
+
+type jsonResponder struct {
+	producer func() (interface{}, error)
+}
+
+func (j jsonResponder) ServeHTTP(w http.ResponseWriter, r *http.Request) {
+	res, err := j.producer()
+	if err != nil {
+		w.WriteHeader(http.StatusInternalServerError)
+		io.WriteString(w, err.Error())
+		return
+	}
+	var buf bytes.Buffer
+	err = json.NewEncoder(&buf).Encode(res)
+	if err != nil {
+		w.WriteHeader(http.StatusInternalServerError)
+		io.WriteString(w, err.Error())
+	} else {
+		io.Copy(w, &buf)
+	}
+}
+
+type jsonDecoder = func(interface{}) error
+
+type jsonRequestResponder struct {
+	producer func(decoder jsonDecoder) (interface{}, error)
+}
+
+func (j jsonRequestResponder) ServeHTTP(w http.ResponseWriter, r *http.Request) {
+	var decodeError error
+	decoder := func(i interface{}) error {
+		err := json.NewDecoder(r.Body).Decode(&i)
+		decodeError = err
+		return err
+	}
+	res, producerErr := j.producer(decoder)
+
+	//If we had a decode error ignore output of producer and return error
+	if decodeError != nil {
+		w.WriteHeader(http.StatusBadRequest)
+		io.WriteString(w, decodeError.Error())
+		return
+	}
+	if producerErr != nil {
+		w.WriteHeader(http.StatusInternalServerError)
+		io.WriteString(w, producerErr.Error())
+		return
+	}
+
+	var buf bytes.Buffer
+	encodeErr := json.NewEncoder(&buf).Encode(res)
+	if encodeErr != nil {
+		w.WriteHeader(http.StatusInternalServerError)
+		io.WriteString(w, encodeErr.Error())
+	} else {
+		io.Copy(w, &buf)
+	}
+}
+
+type requestLogger struct {
+	log         logger.Logger
+	handler     http.Handler
+	handlerFunc http.HandlerFunc
+}
+
+func (l requestLogger) ServeHTTP(w http.ResponseWriter, r *http.Request) {
+	log := l.log.WithField("method", r.Method).WithField("url", r.URL)
+	log.Debug("start")
+	promControl.requestBegin.WithLabelValues(r.URL.Path).Inc()
+	defer prometheus.NewTimer(promControl.requestFinished.WithLabelValues(r.URL.Path)).ObserveDuration()
+	if l.handlerFunc != nil {
+		l.handlerFunc(w, r)
+	} else if l.handler != nil {
+		l.handler.ServeHTTP(w, r)
+	} else {
+		log.Error("no handler or handlerFunc configured")
+	}
+	log.Debug("finish")
+}

daemon/daemon.go

@@ -0,0 +1,223 @@
+package daemon
+
+import (
+	"context"
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/job"
+	"github.com/zrepl/zrepl/daemon/job/reset"
+	"github.com/zrepl/zrepl/daemon/job/wakeup"
+	"github.com/zrepl/zrepl/daemon/logging"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/version"
+	"os"
+	"os/signal"
+	"strings"
+	"sync"
+	"syscall"
+	"time"
+)
+
+func Run(conf *config.Config) error {
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	defer cancel()
+	sigChan := make(chan os.Signal, 1)
+	signal.Notify(sigChan, syscall.SIGINT, syscall.SIGTERM)
+	go func() {
+		<-sigChan
+		cancel()
+	}()
+
+	outlets, err := logging.OutletsFromConfig(*conf.Global.Logging)
+	if err != nil {
+		return errors.Wrap(err, "cannot build logging from config")
+	}
+
+	confJobs, err := job.JobsFromConfig(conf)
+	if err != nil {
+		return errors.Wrap(err, "cannot build jobs from config")
+	}
+
+	log := logger.NewLogger(outlets, 1*time.Second)
+	log.Info(version.NewZreplVersionInformation().String())
+
+	for _, job := range confJobs {
+		if IsInternalJobName(job.Name()) {
+			panic(fmt.Sprintf("internal job name used for config job '%s'", job.Name())) //FIXME
+		}
+	}
+
+	ctx = job.WithLogger(ctx, log)
+
+	jobs := newJobs()
+
+	// start control socket
+	controlJob, err := newControlJob(conf.Global.Control.SockPath, jobs)
+	if err != nil {
+		panic(err) // FIXME
+	}
+	jobs.start(ctx, controlJob, true)
+
+	for i, jc := range conf.Global.Monitoring {
+		var (
+			job job.Job
+			err error
+		)
+		switch v := jc.Ret.(type) {
+		case *config.PrometheusMonitoring:
+			job, err = newPrometheusJobFromConfig(v)
+		default:
+			return errors.Errorf("unknown monitoring job #%d (type %T)", i, v)
+		}
+		if err != nil {
+			return errors.Wrapf(err,"cannot build monitorin gjob #%d", i)
+		}
+		jobs.start(ctx, job, true)
+	}
+
+
+	log.Info("starting daemon")
+
+	// start regular jobs
+	for _, j := range confJobs {
+		jobs.start(ctx, j, false)
+	}
+
+	select {
+	case <-jobs.wait():
+		log.Info("all jobs finished")
+	case <-ctx.Done():
+		log.WithError(ctx.Err()).Info("context finished")
+	}
+	log.Info("daemon exiting")
+	return nil
+}
+
+type jobs struct {
+	wg sync.WaitGroup
+
+	// m protects all fields below it
+	m       sync.RWMutex
+	wakeups map[string]wakeup.Func // by Job.Name
+	resets map[string]reset.Func // by Job.Name
+	jobs    map[string]job.Job
+}
+
+func newJobs() *jobs {
+	return &jobs{
+		wakeups: make(map[string]wakeup.Func),
+		resets:  make(map[string]reset.Func),
+		jobs:    make(map[string]job.Job),
+	}
+}
+
+const (
+	logJobField    string = "job"
+	logTaskField   string = "task"
+	logSubsysField string = "subsystem"
+)
+
+func (s *jobs) wait() <-chan struct{} {
+	ch := make(chan struct{})
+	go func() {
+		s.wg.Wait()
+	}()
+	return ch
+}
+
+func (s *jobs) status() map[string]*job.Status {
+	s.m.RLock()
+	defer s.m.RUnlock()
+
+	type res struct {
+		name   string
+		status *job.Status
+	}
+	var wg sync.WaitGroup
+	c := make(chan res, len(s.jobs))
+	for name, j := range s.jobs {
+		wg.Add(1)
+		go func(name string, j job.Job) {
+			defer wg.Done()
+			c <- res{name: name, status: j.Status()}
+		}(name, j)
+	}
+	wg.Wait()
+	close(c)
+	ret := make(map[string]*job.Status, len(s.jobs))
+	for res := range c {
+		ret[res.name] = res.status
+	}
+	return ret
+}
+
+func (s *jobs) wakeup(job string) error {
+	s.m.RLock()
+	defer s.m.RUnlock()
+
+	wu, ok := s.wakeups[job]
+	if !ok {
+		return errors.Errorf("Job %s does not exist", job)
+	}
+	return wu()
+}
+
+func (s *jobs) reset(job string) error {
+	s.m.RLock()
+	defer s.m.RUnlock()
+
+	wu, ok := s.resets[job]
+	if !ok {
+		return errors.Errorf("Job %s does not exist", job)
+	}
+	return wu()
+}
+
+const (
+	jobNamePrometheus = "_prometheus"
+	jobNameControl    = "_control"
+)
+
+func IsInternalJobName(s string) bool {
+	return strings.HasPrefix(s, "_")
+}
+
+func (s *jobs) start(ctx context.Context, j job.Job, internal bool) {
+	s.m.Lock()
+	defer s.m.Unlock()
+
+	jobLog := job.GetLogger(ctx).
+		WithField(logJobField, j.Name()).
+		WithOutlet(newPrometheusLogOutlet(j.Name()), logger.Debug)
+	jobName := j.Name()
+	if !internal && IsInternalJobName(jobName) {
+		panic(fmt.Sprintf("internal job name used for non-internal job %s", jobName))
+	}
+	if internal && !IsInternalJobName(jobName) {
+		panic(fmt.Sprintf("internal job does not use internal job name %s", jobName))
+	}
+	if _, ok := s.jobs[jobName]; ok {
+		panic(fmt.Sprintf("duplicate job name %s", jobName))
+	}
+
+	j.RegisterMetrics(prometheus.DefaultRegisterer)
+
+	s.jobs[jobName] = j
+	ctx = job.WithLogger(ctx, jobLog)
+	ctx, wakeup := wakeup.Context(ctx)
+	ctx, resetFunc := reset.Context(ctx)
+	s.wakeups[jobName] = wakeup
+	s.resets[jobName] = resetFunc
+
+	s.wg.Add(1)
+	go func() {
+		defer s.wg.Done()
+		jobLog.Info("starting job")
+		defer jobLog.Info("job exited")
+		j.Run(ctx)
+	}()
+}

daemon/filters/fsmapfilter.go

@@ -0,0 +1,273 @@
+package filters
+
+import (
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/endpoint"
+	"github.com/zrepl/zrepl/zfs"
+	"strings"
+)
+
+type DatasetMapFilter struct {
+	entries []datasetMapFilterEntry
+
+	// if set, only valid filter entries can be added using Add()
+	// and Map() will always return an error
+	filterMode bool
+}
+
+type datasetMapFilterEntry struct {
+	path *zfs.DatasetPath
+	// the mapping. since this datastructure acts as both mapping and filter
+	// we have to convert it to the desired rep dynamically
+	mapping      string
+	subtreeMatch bool
+}
+
+func NewDatasetMapFilter(capacity int, filterMode bool) *DatasetMapFilter {
+	return &DatasetMapFilter{
+		entries:    make([]datasetMapFilterEntry, 0, capacity),
+		filterMode: filterMode,
+	}
+}
+
+func (m *DatasetMapFilter) Add(pathPattern, mapping string) (err error) {
+
+	if m.filterMode {
+		if _, err = m.parseDatasetFilterResult(mapping); err != nil {
+			return
+		}
+	}
+
+	// assert path glob adheres to spec
+	const SUBTREE_PATTERN string = "<"
+	patternCount := strings.Count(pathPattern, SUBTREE_PATTERN)
+	switch {
+	case patternCount > 1:
+	case patternCount == 1 && !strings.HasSuffix(pathPattern, SUBTREE_PATTERN):
+		err = fmt.Errorf("pattern invalid: only one '<' at end of string allowed")
+		return
+	}
+
+	pathStr := strings.TrimSuffix(pathPattern, SUBTREE_PATTERN)
+	path, err := zfs.NewDatasetPath(pathStr)
+	if err != nil {
+		return fmt.Errorf("pattern is not a dataset path: %s", err)
+	}
+
+	entry := datasetMapFilterEntry{
+		path:         path,
+		mapping:      mapping,
+		subtreeMatch: patternCount > 0,
+	}
+	m.entries = append(m.entries, entry)
+	return
+
+}
+
+// find the most specific prefix mapping we have
+//
+// longer prefix wins over shorter prefix, direct wins over glob
+func (m DatasetMapFilter) mostSpecificPrefixMapping(path *zfs.DatasetPath) (idx int, found bool) {
+	lcp, lcp_entry_idx := -1, -1
+	direct_idx := -1
+	for e := range m.entries {
+		entry := m.entries[e]
+		ep := m.entries[e].path
+		lep := ep.Length()
+
+		switch {
+		case !entry.subtreeMatch && ep.Equal(path):
+			direct_idx = e
+			continue
+		case entry.subtreeMatch && path.HasPrefix(ep) && lep > lcp:
+			lcp = lep
+			lcp_entry_idx = e
+		default:
+			continue
+		}
+	}
+
+	if lcp_entry_idx >= 0 || direct_idx >= 0 {
+		found = true
+		switch {
+		case direct_idx >= 0:
+			idx = direct_idx
+		case lcp_entry_idx >= 0:
+			idx = lcp_entry_idx
+		}
+	}
+	return
+}
+
+// Returns target == nil if there is no mapping
+func (m DatasetMapFilter) Map(source *zfs.DatasetPath) (target *zfs.DatasetPath, err error) {
+
+	if m.filterMode {
+		err = fmt.Errorf("using a filter for mapping simply does not work")
+		return
+	}
+
+	mi, hasMapping := m.mostSpecificPrefixMapping(source)
+	if !hasMapping {
+		return nil, nil
+	}
+	me := m.entries[mi]
+
+	if me.mapping == "" {
+		// Special case treatment: 'foo/bar<' => ''
+		if !me.subtreeMatch {
+			return nil, fmt.Errorf("mapping to '' must be a subtree match")
+		}
+		// ok...
+	} else {
+		if strings.HasPrefix("!", me.mapping) {
+			// reject mapping
+			return nil, nil
+		}
+	}
+
+	target, err = zfs.NewDatasetPath(me.mapping)
+	if err != nil {
+		err = fmt.Errorf("mapping target is not a dataset path: %s", err)
+		return
+	}
+	if me.subtreeMatch {
+		// strip common prefix ('<' wildcards are no special case here)
+		extendComps := source.Copy()
+		extendComps.TrimPrefix(me.path)
+		target.Extend(extendComps)
+	}
+	return
+}
+
+func (m DatasetMapFilter) Filter(p *zfs.DatasetPath) (pass bool, err error) {
+
+	if !m.filterMode {
+		err = fmt.Errorf("using a mapping as a filter does not work")
+		return
+	}
+
+	mi, hasMapping := m.mostSpecificPrefixMapping(p)
+	if !hasMapping {
+		pass = false
+		return
+	}
+	me := m.entries[mi]
+	pass, err = m.parseDatasetFilterResult(me.mapping)
+	return
+}
+
+// Construct a new filter-only DatasetMapFilter from a mapping
+// The new filter allows excactly those paths that were not forbidden by the mapping.
+func (m DatasetMapFilter) InvertedFilter() (inv *DatasetMapFilter, err error) {
+
+	if m.filterMode {
+		err = errors.Errorf("can only invert mappings")
+		return
+	}
+
+	inv = &DatasetMapFilter{
+		make([]datasetMapFilterEntry, len(m.entries)),
+		true,
+	}
+
+	for i, e := range m.entries {
+		inv.entries[i].path, err = zfs.NewDatasetPath(e.mapping)
+		if err != nil {
+			err = errors.Wrapf(err, "mapping cannot be inverted: '%s' is not a dataset path", e.mapping)
+			return
+		}
+		inv.entries[i].mapping = MapFilterResultOk
+		inv.entries[i].subtreeMatch = e.subtreeMatch
+	}
+
+	return inv, nil
+}
+
+// FIXME investigate whether we can support more...
+func (m DatasetMapFilter) Invert() (endpoint.FSMap, error) {
+
+	if m.filterMode {
+		return nil, errors.Errorf("can only invert mappings")
+	}
+
+	if len(m.entries) != 1 {
+		return nil, errors.Errorf("inversion of complicated mappings is not implemented") // FIXME
+	}
+
+	e := m.entries[0]
+
+	inv := &DatasetMapFilter{
+		make([]datasetMapFilterEntry, len(m.entries)),
+		false,
+	}
+	mp, err := zfs.NewDatasetPath(e.mapping)
+	if err != nil {
+		return nil, err
+	}
+
+	inv.entries[0] = datasetMapFilterEntry{
+		path:         mp,
+		mapping:      e.path.ToString(),
+		subtreeMatch: e.subtreeMatch,
+	}
+
+	return inv, nil
+}
+
+// Creates a new DatasetMapFilter in filter mode from a mapping
+// All accepting mapping results are mapped to accepting filter results
+// All rejecting mapping results are mapped to rejecting filter results
+func (m DatasetMapFilter) AsFilter() endpoint.FSFilter {
+
+	f := &DatasetMapFilter{
+		make([]datasetMapFilterEntry, len(m.entries)),
+		true,
+	}
+
+	for i, e := range m.entries {
+		var newe datasetMapFilterEntry = e
+		if strings.HasPrefix(newe.mapping, "!") {
+			newe.mapping = MapFilterResultOmit
+		} else {
+			newe.mapping = MapFilterResultOk
+		}
+		f.entries[i] = newe
+	}
+
+	return f
+}
+
+const (
+	MapFilterResultOk   string = "ok"
+	MapFilterResultOmit string = "!"
+)
+
+// Parse a dataset filter result
+func (m DatasetMapFilter) parseDatasetFilterResult(result string) (pass bool, err error) {
+	l := strings.ToLower(result)
+	if l == MapFilterResultOk {
+		return true, nil
+	}
+	if l == MapFilterResultOmit {
+		return false, nil
+	}
+	return false, fmt.Errorf("'%s' is not a valid filter result", result)
+}
+
+func DatasetMapFilterFromConfig(in map[string]bool) (f *DatasetMapFilter, err error) {
+
+	f = NewDatasetMapFilter(len(in), true)
+	for pathPattern, accept := range in {
+		mapping := MapFilterResultOmit
+		if accept {
+			mapping = MapFilterResultOk
+		}
+		if err = f.Add(pathPattern, mapping); err != nil {
+			err = fmt.Errorf("invalid mapping entry ['%s':'%s']: %s", pathPattern, mapping, err)
+			return
+		}
+	}
+	return
+}

daemon/filters/fsvfilter.go

@@ -0,0 +1,41 @@
+package filters
+
+import (
+	"github.com/zrepl/zrepl/zfs"
+	"strings"
+)
+
+type AnyFSVFilter struct{}
+
+func NewAnyFSVFilter() AnyFSVFilter {
+	return AnyFSVFilter{}
+}
+
+var _ zfs.FilesystemVersionFilter = AnyFSVFilter{}
+
+func (AnyFSVFilter) Filter(t zfs.VersionType, name string) (accept bool, err error) {
+	return true, nil
+}
+
+
+type PrefixFilter struct {
+	prefix    string
+	fstype    zfs.VersionType
+	fstypeSet bool // optionals anyone?
+}
+
+var _ zfs.FilesystemVersionFilter = &PrefixFilter{}
+
+func NewPrefixFilter(prefix string) *PrefixFilter {
+	return &PrefixFilter{prefix: prefix}
+}
+
+func NewTypedPrefixFilter(prefix string, versionType zfs.VersionType) *PrefixFilter {
+	return &PrefixFilter{prefix, versionType, true}
+}
+
+func (f *PrefixFilter) Filter(t zfs.VersionType, name string) (accept bool, err error) {
+	fstypeMatches := (!f.fstypeSet || t == f.fstype)
+	prefixMatches := strings.HasPrefix(name, f.prefix)
+	return fstypeMatches && prefixMatches, nil
+}

daemon/job/active.go

@@ -0,0 +1,422 @@
+package job
+
+import (
+	"context"
+	"github.com/pkg/errors"
+	"github.com/problame/go-streamrpc"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/filters"
+	"github.com/zrepl/zrepl/daemon/job/reset"
+	"github.com/zrepl/zrepl/daemon/job/wakeup"
+	"github.com/zrepl/zrepl/daemon/logging"
+	"github.com/zrepl/zrepl/daemon/pruner"
+	"github.com/zrepl/zrepl/daemon/snapper"
+	"github.com/zrepl/zrepl/daemon/transport/connecter"
+	"github.com/zrepl/zrepl/endpoint"
+	"github.com/zrepl/zrepl/replication"
+	"github.com/zrepl/zrepl/util/envconst"
+	"github.com/zrepl/zrepl/zfs"
+	"sync"
+	"time"
+)
+
+type ActiveSide struct {
+	mode          activeMode
+	name          string
+	clientFactory *connecter.ClientFactory
+
+	prunerFactory *pruner.PrunerFactory
+
+
+	promRepStateSecs *prometheus.HistogramVec // labels: state
+	promPruneSecs *prometheus.HistogramVec // labels: prune_side
+	promBytesReplicated *prometheus.CounterVec // labels: filesystem
+
+	tasksMtx sync.Mutex
+	tasks    activeSideTasks
+}
+
+
+//go:generate enumer -type=ActiveSideState
+type ActiveSideState int
+
+const (
+	ActiveSideReplicating ActiveSideState = 1 << iota
+	ActiveSidePruneSender
+	ActiveSidePruneReceiver
+	ActiveSideDone // also errors
+)
+
+
+type activeSideTasks struct {
+	state ActiveSideState
+
+	// valid for state ActiveSideReplicating, ActiveSidePruneSender, ActiveSidePruneReceiver, ActiveSideDone
+	replication *replication.Replication
+	replicationCancel context.CancelFunc
+
+	// valid for state ActiveSidePruneSender, ActiveSidePruneReceiver, ActiveSideDone
+	prunerSender, prunerReceiver *pruner.Pruner
+
+	// valid for state ActiveSidePruneReceiver, ActiveSideDone
+	prunerSenderCancel, prunerReceiverCancel context.CancelFunc
+}
+
+func (a *ActiveSide) updateTasks(u func(*activeSideTasks)) activeSideTasks {
+	a.tasksMtx.Lock()
+	defer a.tasksMtx.Unlock()
+	var copy activeSideTasks
+	copy = a.tasks
+	if u == nil {
+		return copy
+	}
+	u(&copy)
+	a.tasks = copy
+	return copy
+}
+
+type activeMode interface {
+	SenderReceiver(client *streamrpc.Client) (replication.Sender, replication.Receiver, error)
+	Type() Type
+	RunPeriodic(ctx context.Context, wakeUpCommon chan<- struct{})
+}
+
+type modePush struct {
+	fsfilter         endpoint.FSFilter
+	snapper *snapper.PeriodicOrManual
+}
+
+func (m *modePush) SenderReceiver(client *streamrpc.Client) (replication.Sender, replication.Receiver, error) {
+	sender := endpoint.NewSender(m.fsfilter)
+	receiver := endpoint.NewRemote(client)
+	return sender, receiver, nil
+}
+
+func (m *modePush) Type() Type { return TypePush }
+
+func (m *modePush) RunPeriodic(ctx context.Context, wakeUpCommon chan <- struct{}) {
+	m.snapper.Run(ctx, wakeUpCommon)
+}
+
+
+func modePushFromConfig(g *config.Global, in *config.PushJob) (*modePush, error) {
+	m := &modePush{}
+	fsf, err := filters.DatasetMapFilterFromConfig(in.Filesystems)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannnot build filesystem filter")
+	}
+	m.fsfilter = fsf
+
+	if m.snapper, err = snapper.FromConfig(g, fsf, in.Snapshotting); err != nil {
+		return nil, errors.Wrap(err, "cannot build snapper")
+	}
+
+	return m, nil
+}
+
+type modePull struct {
+	rootFS   *zfs.DatasetPath
+	interval time.Duration
+}
+
+func (m *modePull) SenderReceiver(client *streamrpc.Client) (replication.Sender, replication.Receiver, error) {
+	sender := endpoint.NewRemote(client)
+	receiver, err := endpoint.NewReceiver(m.rootFS)
+	return sender, receiver, err
+}
+
+func (*modePull) Type() Type { return TypePull }
+
+func (m *modePull) RunPeriodic(ctx context.Context, wakeUpCommon chan<- struct{}) {
+	t := time.NewTicker(m.interval)
+	defer t.Stop()
+	for {
+		select {
+		case <-t.C:
+			select {
+			case wakeUpCommon <- struct{}{}:
+			default:
+				GetLogger(ctx).
+					WithField("pull_interval", m.interval).
+					Warn("pull job took longer than pull interval")
+				wakeUpCommon <- struct{}{} // block anyways, to queue up the wakeup
+			}
+		case <-ctx.Done():
+			return
+		}
+	}
+}
+
+func modePullFromConfig(g *config.Global, in *config.PullJob) (m *modePull, err error) {
+	m = &modePull{}
+	if in.Interval <= 0 {
+		return nil, errors.New("interval must be positive")
+	}
+	m.interval = in.Interval
+
+	m.rootFS, err = zfs.NewDatasetPath(in.RootFS)
+	if err != nil {
+		return nil, errors.New("RootFS is not a valid zfs filesystem path")
+	}
+	if m.rootFS.Length() <= 0 {
+		return nil, errors.New("RootFS must not be empty") // duplicates error check of receiver
+	}
+
+	return m, nil
+}
+
+func activeSide(g *config.Global, in *config.ActiveJob, mode activeMode) (j *ActiveSide, err error) {
+
+	j = &ActiveSide{mode: mode}
+	j.name = in.Name
+	j.promRepStateSecs = prometheus.NewHistogramVec(prometheus.HistogramOpts{
+		Namespace:   "zrepl",
+		Subsystem:   "replication",
+		Name:        "state_time",
+		Help:        "seconds spent during replication",
+		ConstLabels: prometheus.Labels{"zrepl_job":j.name},
+	}, []string{"state"})
+	j.promBytesReplicated = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Namespace:   "zrepl",
+		Subsystem:   "replication",
+		Name:        "bytes_replicated",
+		Help:        "number of bytes replicated from sender to receiver per filesystem",
+		ConstLabels: prometheus.Labels{"zrepl_job":j.name},
+	}, []string{"filesystem"})
+
+	j.clientFactory, err = connecter.FromConfig(g, in.Connect)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot build client")
+	}
+
+	j.promPruneSecs = prometheus.NewHistogramVec(prometheus.HistogramOpts{
+		Namespace:   "zrepl",
+		Subsystem:   "pruning",
+		Name:        "time",
+		Help:        "seconds spent in pruner",
+		ConstLabels: prometheus.Labels{"zrepl_job":j.name},
+	}, []string{"prune_side"})
+	j.prunerFactory, err = pruner.NewPrunerFactory(in.Pruning, j.promPruneSecs)
+	if err != nil {
+		return nil, err
+	}
+
+	return j, nil
+}
+
+func (j *ActiveSide) RegisterMetrics(registerer prometheus.Registerer) {
+	registerer.MustRegister(j.promRepStateSecs)
+	registerer.MustRegister(j.promPruneSecs)
+	registerer.MustRegister(j.promBytesReplicated)
+}
+
+func (j *ActiveSide) Name() string { return j.name }
+
+type ActiveSideStatus struct {
+	Replication *replication.Report
+	PruningSender, PruningReceiver *pruner.Report
+}
+
+func (j *ActiveSide) Status() *Status {
+	tasks := j.updateTasks(nil)
+
+	s := &ActiveSideStatus{}
+	t := j.mode.Type()
+	if tasks.replication != nil {
+		s.Replication = tasks.replication.Report()
+	}
+	if tasks.prunerSender != nil {
+		s.PruningSender = tasks.prunerSender.Report()
+	}
+	if tasks.prunerReceiver != nil {
+		s.PruningReceiver = tasks.prunerReceiver.Report()
+	}
+	return &Status{Type: t, JobSpecific: s}
+}
+
+func (j *ActiveSide) Run(ctx context.Context) {
+	log := GetLogger(ctx)
+	ctx = logging.WithSubsystemLoggers(ctx, log)
+
+	defer log.Info("job exiting")
+
+	periodicDone := make(chan struct{})
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+	go j.mode.RunPeriodic(ctx, periodicDone)
+
+	invocationCount := 0
+outer:
+	for {
+		log.Info("wait for wakeups")
+		select {
+		case <-ctx.Done():
+			log.WithError(ctx.Err()).Info("context")
+			break outer
+
+		case <-wakeup.Wait(ctx):
+		case <-periodicDone:
+		}
+		invocationCount++
+		invLog := log.WithField("invocation", invocationCount)
+		j.do(WithLogger(ctx, invLog))
+	}
+}
+
+func (j *ActiveSide) do(ctx context.Context) {
+
+	log := GetLogger(ctx)
+	ctx = logging.WithSubsystemLoggers(ctx, log)
+
+	// allow cancellation of an invocation (this function)
+	ctx, cancelThisRun := context.WithCancel(ctx)
+	defer cancelThisRun()
+	go func() {
+		select {
+		case <-reset.Wait(ctx):
+			log.Info("reset received, cancelling current invocation")
+			cancelThisRun()
+		case <-ctx.Done():
+		}
+	}()
+
+	// The code after this watchdog goroutine is sequential and transitions the state from
+	//   ActiveSideReplicating -> ActiveSidePruneSender -> ActiveSidePruneReceiver -> ActiveSideDone
+	// If any of those sequential tasks 'gets stuck' (livelock, no progress), the watchdog will eventually
+	// cancel its context.
+	// If the task is written to support context cancellation, it will return immediately (in permanent error state),
+	// and the sequential code above transitions to the next state.
+	go func() {
+
+		wdto := envconst.Duration("ZREPL_JOB_WATCHDOG_TIMEOUT", 10*time.Minute)
+		jitter := envconst.Duration("ZREPL_JOB_WATCHDOG_JITTER", 1*time.Second)
+		// shadowing!
+		log := log.WithField("watchdog_timeout", wdto.String())
+
+		log.Debug("starting watchdog")
+		defer log.Debug("watchdog stopped")
+
+		t := time.NewTicker(wdto)
+		defer t.Stop()
+
+		for {
+			select {
+			case <-ctx.Done():
+				return
+			case <-t.C: // fall
+			}
+
+			j.updateTasks(func(tasks *activeSideTasks) {
+				// Since cancelling a task will cause the sequential code to transition to the next state immediately,
+				// we cannot check for its progress right then (no fallthrough).
+				// Instead, we return (not continue because we are in a closure) and give the new state another
+				// ZREPL_JOB_WATCHDOG_TIMEOUT interval to try make some progress.
+
+				log.WithField("state", tasks.state).Debug("watchdog firing")
+
+				const WATCHDOG_ENVCONST_NOTICE = " (adjust ZREPL_JOB_WATCHDOG_TIMEOUT env variable if inappropriate)"
+
+				switch tasks.state {
+				case ActiveSideReplicating:
+					log.WithField("replication_progress", tasks.replication.Progress.String()).
+						Debug("check replication progress")
+					if tasks.replication.Progress.CheckTimeout(wdto, jitter) {
+						log.Error("replication did not make progress, cancelling" + WATCHDOG_ENVCONST_NOTICE)
+						tasks.replicationCancel()
+						return
+					}
+				case ActiveSidePruneSender:
+					log.WithField("prune_sender_progress", tasks.replication.Progress.String()).
+						Debug("check pruner_sender progress")
+					if tasks.prunerSender.Progress.CheckTimeout(wdto, jitter) {
+						log.Error("pruner_sender did not make progress, cancelling" + WATCHDOG_ENVCONST_NOTICE)
+						tasks.prunerSenderCancel()
+						return
+					}
+				case ActiveSidePruneReceiver:
+					log.WithField("prune_receiver_progress", tasks.replication.Progress.String()).
+						Debug("check pruner_receiver progress")
+					if tasks.prunerReceiver.Progress.CheckTimeout(wdto, jitter) {
+						log.Error("pruner_receiver did not make progress, cancelling" + WATCHDOG_ENVCONST_NOTICE)
+						tasks.prunerReceiverCancel()
+						return
+					}
+				case ActiveSideDone:
+					// ignore, ctx will be Done() in a few milliseconds and the watchdog will exit
+				default:
+					log.WithField("state", tasks.state).
+						Error("watchdog implementation error: unknown active side state")
+				}
+			})
+
+		}
+	}()
+
+	client, err := j.clientFactory.NewClient()
+	if err != nil {
+		log.WithError(err).Error("factory cannot instantiate streamrpc client")
+	}
+	defer client.Close(ctx)
+
+	sender, receiver, err := j.mode.SenderReceiver(client)
+
+	{
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+		ctx, repCancel := context.WithCancel(ctx)
+		tasks := j.updateTasks(func(tasks *activeSideTasks) {
+			// reset it
+			*tasks = activeSideTasks{}
+			tasks.replicationCancel = repCancel
+			tasks.replication = replication.NewReplication(j.promRepStateSecs, j.promBytesReplicated)
+			tasks.state = ActiveSideReplicating
+		})
+		log.Info("start replication")
+		tasks.replication.Drive(ctx, sender, receiver)
+		repCancel() // always cancel to free up context resources
+	}
+
+	{
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+		ctx, senderCancel := context.WithCancel(ctx)
+		tasks := j.updateTasks(func(tasks *activeSideTasks) {
+			tasks.prunerSender = j.prunerFactory.BuildSenderPruner(ctx, sender, sender)
+			tasks.prunerSenderCancel = senderCancel
+			tasks.state = ActiveSidePruneSender
+		})
+		log.Info("start pruning sender")
+		tasks.prunerSender.Prune()
+		log.Info("finished pruning sender")
+		senderCancel()
+	}
+	{
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+		ctx, receiverCancel := context.WithCancel(ctx)
+		tasks := j.updateTasks(func(tasks *activeSideTasks) {
+			tasks.prunerReceiver = j.prunerFactory.BuildReceiverPruner(ctx, receiver, sender)
+			tasks.prunerReceiverCancel = receiverCancel
+			tasks.state = ActiveSidePruneReceiver
+		})
+		log.Info("start pruning receiver")
+		tasks.prunerReceiver.Prune()
+		log.Info("finished pruning receiver")
+		receiverCancel()
+	}
+
+	j.updateTasks(func(tasks *activeSideTasks) {
+		tasks.state = ActiveSideDone
+	})
+
+}

daemon/job/activesidestate_enumer.go

@@ -0,0 +1,66 @@
+// Code generated by "enumer -type=ActiveSideState"; DO NOT EDIT.
+
+package job
+
+import (
+	"fmt"
+)
+
+const (
+	_ActiveSideStateName_0 = "ActiveSideReplicatingActiveSidePruneSender"
+	_ActiveSideStateName_1 = "ActiveSidePruneReceiver"
+	_ActiveSideStateName_2 = "ActiveSideDone"
+)
+
+var (
+	_ActiveSideStateIndex_0 = [...]uint8{0, 21, 42}
+	_ActiveSideStateIndex_1 = [...]uint8{0, 23}
+	_ActiveSideStateIndex_2 = [...]uint8{0, 14}
+)
+
+func (i ActiveSideState) String() string {
+	switch {
+	case 1 <= i && i <= 2:
+		i -= 1
+		return _ActiveSideStateName_0[_ActiveSideStateIndex_0[i]:_ActiveSideStateIndex_0[i+1]]
+	case i == 4:
+		return _ActiveSideStateName_1
+	case i == 8:
+		return _ActiveSideStateName_2
+	default:
+		return fmt.Sprintf("ActiveSideState(%d)", i)
+	}
+}
+
+var _ActiveSideStateValues = []ActiveSideState{1, 2, 4, 8}
+
+var _ActiveSideStateNameToValueMap = map[string]ActiveSideState{
+	_ActiveSideStateName_0[0:21]:  1,
+	_ActiveSideStateName_0[21:42]: 2,
+	_ActiveSideStateName_1[0:23]:  4,
+	_ActiveSideStateName_2[0:14]:  8,
+}
+
+// ActiveSideStateString retrieves an enum value from the enum constants string name.
+// Throws an error if the param is not part of the enum.
+func ActiveSideStateString(s string) (ActiveSideState, error) {
+	if val, ok := _ActiveSideStateNameToValueMap[s]; ok {
+		return val, nil
+	}
+	return 0, fmt.Errorf("%s does not belong to ActiveSideState values", s)
+}
+
+// ActiveSideStateValues returns all values of the enum
+func ActiveSideStateValues() []ActiveSideState {
+	return _ActiveSideStateValues
+}
+
+// IsAActiveSideState returns "true" if the value is listed in the enum definition. "false" otherwise
+func (i ActiveSideState) IsAActiveSideState() bool {
+	for _, v := range _ActiveSideStateValues {
+		if i == v {
+			return true
+		}
+	}
+	return false
+}

daemon/job/build_jobs.go

@@ -0,0 +1,71 @@
+package job
+
+import (
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/config"
+)
+
+func JobsFromConfig(c *config.Config) ([]Job, error) {
+	js := make([]Job, len(c.Jobs))
+	for i := range c.Jobs {
+		j, err := buildJob(c.Global, c.Jobs[i])
+		if err != nil {
+			return nil, err
+		}
+		if j == nil || j.Name() == "" {
+			panic(fmt.Sprintf("implementation error: job builder returned nil job type %T", c.Jobs[i].Ret))
+		}
+		js[i] = j
+	}
+	return js, nil
+}
+
+func buildJob(c *config.Global, in config.JobEnum) (j Job, err error) {
+	cannotBuildJob := func(e error, name string) (Job, error) {
+		return nil, errors.Wrapf(e, "cannot build job %q", name)
+	}
+	// FIXME prettify this
+	switch v := in.Ret.(type) {
+	case *config.SinkJob:
+		m, err := modeSinkFromConfig(c, v)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+		j, err = passiveSideFromConfig(c, &v.PassiveJob, m)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+	case *config.SourceJob:
+		m, err := modeSourceFromConfig(c, v)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+		j, err = passiveSideFromConfig(c, &v.PassiveJob, m)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+	case *config.PushJob:
+		m, err := modePushFromConfig(c, v)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+		j, err = activeSide(c, &v.ActiveJob, m)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+	case *config.PullJob:
+		m, err := modePullFromConfig(c, v)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+		j, err = activeSide(c, &v.ActiveJob, m)
+		if err != nil {
+			return cannotBuildJob(err, v.Name)
+		}
+	default:
+		panic(fmt.Sprintf("implementation error: unknown job type %T", v))
+	}
+	return j, nil
+
+}

daemon/job/job.go

@@ -0,0 +1,103 @@
+package job
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/logger"
+)
+
+type Logger = logger.Logger
+
+type contextKey int
+
+const (
+	contextKeyLog contextKey = iota
+)
+
+func GetLogger(ctx context.Context) Logger {
+	if l, ok := ctx.Value(contextKeyLog).(Logger); ok {
+		return l
+	}
+	return logger.NewNullLogger()
+}
+
+func WithLogger(ctx context.Context, l Logger) context.Context {
+	return context.WithValue(ctx, contextKeyLog, l)
+}
+
+
+type Job interface {
+	Name() string
+	Run(ctx context.Context)
+	Status() *Status
+	RegisterMetrics(registerer prometheus.Registerer)
+}
+
+type Type string
+
+const (
+	TypeInternal Type = "internal"
+	TypePush Type = "push"
+	TypeSink Type = "sink"
+	TypePull Type  = "pull"
+	TypeSource Type = "source"
+)
+
+type Status struct {
+	Type Type
+	JobSpecific interface{}
+}
+
+func (s *Status) MarshalJSON() ([]byte, error) {
+	typeJson, err := json.Marshal(s.Type)
+	if err != nil {
+		return nil, err
+	}
+	jobJSON, err := json.Marshal(s.JobSpecific)
+	if err != nil {
+		return nil, err
+	}
+	m := map[string]json.RawMessage {
+		"type": typeJson,
+		string(s.Type): jobJSON,
+	}
+	return json.Marshal(m)
+}
+
+func (s *Status) UnmarshalJSON(in []byte) (err error) {
+	var m map[string]json.RawMessage
+	if err := json.Unmarshal(in, &m); err != nil {
+		return err
+	}
+	tJSON, ok := m["type"]
+	if !ok {
+		return fmt.Errorf("field 'type' not found")
+	}
+	if err := json.Unmarshal(tJSON, &s.Type); err != nil {
+		return err
+	}
+	key := string(s.Type)
+	jobJSON, ok := m[key]
+	if !ok {
+		return fmt.Errorf("field '%s', not found", key)
+	}
+	switch s.Type {
+	case TypePull: fallthrough
+	case TypePush:
+		var st ActiveSideStatus
+		err = json.Unmarshal(jobJSON, &st)
+		s.JobSpecific = &st
+	case TypeSource: fallthrough
+	case TypeSink:
+		var st PassiveStatus
+		err = json.Unmarshal(jobJSON, &st)
+		s.JobSpecific = &st
+	case TypeInternal:
+		// internal jobs do not report specifics
+	default:
+		err = fmt.Errorf("unknown job type '%s'", key)
+	}
+	return err
+}

daemon/job/passive.go

@@ -0,0 +1,196 @@
+package job
+
+import (
+	"context"
+	"github.com/pkg/errors"
+	"github.com/problame/go-streamrpc"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/filters"
+	"github.com/zrepl/zrepl/daemon/logging"
+	"github.com/zrepl/zrepl/daemon/transport/serve"
+	"github.com/zrepl/zrepl/daemon/snapper"
+	"github.com/zrepl/zrepl/endpoint"
+	"github.com/zrepl/zrepl/zfs"
+	"path"
+)
+
+type PassiveSide struct {
+	mode passiveMode
+	name     string
+	l        serve.ListenerFactory
+	rpcConf  *streamrpc.ConnConfig
+}
+
+type passiveMode interface {
+	ConnHandleFunc(ctx context.Context, conn serve.AuthenticatedConn) streamrpc.HandlerFunc
+	RunPeriodic(ctx context.Context)
+	Type() Type
+}
+
+type modeSink struct {
+	rootDataset *zfs.DatasetPath
+}
+
+func (m *modeSink) Type() Type { return TypeSink }
+
+func (m *modeSink) ConnHandleFunc(ctx context.Context, conn serve.AuthenticatedConn) streamrpc.HandlerFunc {
+	log := GetLogger(ctx)
+
+	clientRootStr := path.Join(m.rootDataset.ToString(), conn.ClientIdentity())
+	clientRoot, err := zfs.NewDatasetPath(clientRootStr)
+	if err != nil {
+		log.WithError(err).
+			WithField("client_identity", conn.ClientIdentity()).
+			Error("cannot build client filesystem map (client identity must be a valid ZFS FS name")
+	}
+	log.WithField("client_root", clientRoot).Debug("client root")
+
+	local, err := endpoint.NewReceiver(clientRoot)
+	if err != nil {
+		log.WithError(err).Error("unexpected error: cannot convert mapping to filter")
+		return nil
+	}
+
+	h := endpoint.NewHandler(local)
+	return h.Handle
+}
+
+func (m *modeSink) RunPeriodic(_ context.Context) {}
+
+func modeSinkFromConfig(g *config.Global, in *config.SinkJob) (m *modeSink, err error) {
+	m = &modeSink{}
+	m.rootDataset, err = zfs.NewDatasetPath(in.RootFS)
+	if err != nil {
+		return nil, errors.New("root dataset is not a valid zfs filesystem path")
+	}
+	if m.rootDataset.Length() <= 0 {
+		return nil, errors.New("root dataset must not be empty") // duplicates error check of receiver
+	}
+	return m, nil
+}
+
+type modeSource struct {
+	fsfilter zfs.DatasetFilter
+	snapper *snapper.PeriodicOrManual
+}
+
+func modeSourceFromConfig(g *config.Global, in *config.SourceJob) (m *modeSource, err error) {
+	// FIXME exact dedup of modePush
+	m = &modeSource{}
+	fsf, err := filters.DatasetMapFilterFromConfig(in.Filesystems)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannnot build filesystem filter")
+	}
+	m.fsfilter = fsf
+
+	if m.snapper, err = snapper.FromConfig(g, fsf, in.Snapshotting); err != nil {
+		return nil, errors.Wrap(err, "cannot build snapper")
+	}
+
+	return m, nil
+}
+
+func (m *modeSource) Type() Type { return TypeSource }
+
+func (m *modeSource) ConnHandleFunc(ctx context.Context, conn serve.AuthenticatedConn) streamrpc.HandlerFunc {
+	sender := endpoint.NewSender(m.fsfilter)
+	h := endpoint.NewHandler(sender)
+	return h.Handle
+}
+
+func (m *modeSource) RunPeriodic(ctx context.Context) {
+	m.snapper.Run(ctx, nil)
+}
+
+func passiveSideFromConfig(g *config.Global, in *config.PassiveJob, mode passiveMode) (s *PassiveSide, err error) {
+
+	s = &PassiveSide{mode: mode, name: in.Name}
+	if s.l, s.rpcConf, err = serve.FromConfig(g, in.Serve); err != nil {
+		return nil, errors.Wrap(err, "cannot build server")
+	}
+
+	return s, nil
+}
+
+func (j *PassiveSide) Name() string { return j.name }
+
+type PassiveStatus struct {}
+
+func (s *PassiveSide) Status() *Status {
+	return &Status{Type: s.mode.Type()} // FIXME PassiveStatus
+}
+
+func (*PassiveSide) RegisterMetrics(registerer prometheus.Registerer) {}
+
+func (j *PassiveSide) Run(ctx context.Context) {
+
+	log := GetLogger(ctx)
+	defer log.Info("job exiting")
+
+	l, err := j.l.Listen()
+	if err != nil {
+		log.WithError(err).Error("cannot listen")
+		return
+	}
+	defer l.Close()
+
+	{
+		ctx, cancel := context.WithCancel(logging.WithSubsystemLoggers(ctx, log)) // shadowing
+		defer cancel()
+		go j.mode.RunPeriodic(ctx)
+	}
+
+	log.WithField("addr", l.Addr()).Debug("accepting connections")
+	var connId int
+outer:
+	for {
+
+		select {
+		case res := <-accept(ctx, l):
+			if res.err != nil {
+				log.WithError(res.err).Info("accept error")
+				continue
+			}
+			conn := res.conn
+			connId++
+			connLog := log.
+				WithField("connID", connId)
+			connLog.
+				WithField("addr", conn.RemoteAddr()).
+				WithField("client_identity", conn.ClientIdentity()).
+				Info("handling connection")
+			go func() {
+				defer connLog.Info("finished handling connection")
+				defer conn.Close()
+				ctx := logging.WithSubsystemLoggers(ctx, connLog)
+				handleFunc := j.mode.ConnHandleFunc(ctx, conn)
+				if handleFunc == nil {
+					return
+				}
+				if err := streamrpc.ServeConn(ctx, conn, j.rpcConf, handleFunc); err != nil {
+					log.WithError(err).Error("error serving client")
+				}
+			}()
+
+		case <-ctx.Done():
+			break outer
+		}
+
+	}
+
+}
+
+type acceptResult struct {
+	conn serve.AuthenticatedConn
+	err  error
+}
+
+func accept(ctx context.Context, listener serve.AuthenticatedListener) <-chan acceptResult {
+	c := make(chan acceptResult, 1)
+	go func() {
+		conn, err := listener.Accept(ctx)
+		c <- acceptResult{conn, err}
+	}()
+	return c
+}

daemon/job/reset/reset.go

@@ -0,0 +1,35 @@
+package reset
+
+import (
+	"context"
+	"errors"
+)
+
+type contextKey int
+
+const contextKeyReset contextKey = iota
+
+func Wait(ctx context.Context) <-chan struct{} {
+	wc, ok := ctx.Value(contextKeyReset).(chan struct{})
+	if !ok {
+		wc = make(chan struct{})
+	}
+	return wc
+}
+
+type Func func() error
+
+var AlreadyReset = errors.New("already reset")
+
+func Context(ctx context.Context) (context.Context, Func) {
+	wc := make(chan struct{})
+	wuf := func() error {
+		select {
+		case wc <- struct{}{}:
+			return nil
+		default:
+			return AlreadyReset
+		}
+	}
+	return context.WithValue(ctx, contextKeyReset, wc), wuf
+}

daemon/job/wakeup/wakeup.go

@@ -0,0 +1,35 @@
+package wakeup
+
+import (
+	"context"
+	"errors"
+)
+
+type contextKey int
+
+const contextKeyWakeup contextKey = iota
+
+func Wait(ctx context.Context) <-chan struct{} {
+	wc, ok := ctx.Value(contextKeyWakeup).(chan struct{})
+	if !ok {
+		wc = make(chan struct{})
+	}
+	return wc
+}
+
+type Func func() error
+
+var AlreadyWokenUp = errors.New("already woken up")
+
+func Context(ctx context.Context) (context.Context, Func) {
+	wc := make(chan struct{})
+	wuf := func() error {
+		select {
+		case wc <- struct{}{}:
+			return nil
+		default:
+			return AlreadyWokenUp
+		}
+	}
+	return context.WithValue(ctx, contextKeyWakeup, wc), wuf
+}

daemon/logging/adaptors.go

@@ -0,0 +1,32 @@
+package logging
+
+import (
+	"fmt"
+	"github.com/problame/go-streamrpc"
+	"github.com/zrepl/zrepl/logger"
+	"strings"
+)
+
+type streamrpcLogAdaptor = twoClassLogAdaptor
+
+type twoClassLogAdaptor struct {
+	logger.Logger
+}
+
+var _ streamrpc.Logger = twoClassLogAdaptor{}
+
+func (a twoClassLogAdaptor) Errorf(fmtStr string, args ...interface{}) {
+	const errorSuffix = ": %s"
+	if len(args) == 1 {
+		if err, ok := args[0].(error); ok && strings.HasSuffix(fmtStr, errorSuffix) {
+			msg := strings.TrimSuffix(fmtStr, errorSuffix)
+			a.WithError(err).Error(msg)
+			return
+		}
+	}
+	a.Logger.Error(fmt.Sprintf(fmtStr, args...))
+}
+
+func (a twoClassLogAdaptor) Infof(fmtStr string, args ...interface{}) {
+	a.Logger.Debug(fmt.Sprintf(fmtStr, args...))
+}

daemon/logging/build_logging.go

@@ -0,0 +1,205 @@
+package logging
+
+import (
+	"context"
+	"crypto/tls"
+	"crypto/x509"
+	"github.com/mattn/go-isatty"
+	"github.com/pkg/errors"
+	"github.com/problame/go-streamrpc"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/pruner"
+	"github.com/zrepl/zrepl/endpoint"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/replication"
+	"github.com/zrepl/zrepl/tlsconf"
+	"os"
+	"github.com/zrepl/zrepl/daemon/snapper"
+	"github.com/zrepl/zrepl/daemon/transport/serve"
+)
+
+func OutletsFromConfig(in config.LoggingOutletEnumList) (*logger.Outlets, error) {
+
+	outlets := logger.NewOutlets()
+
+	if len(in) == 0 {
+		// Default config
+		out := WriterOutlet{&HumanFormatter{}, os.Stdout}
+		outlets.Add(out, logger.Warn)
+		return outlets, nil
+	}
+
+	var syslogOutlets, stdoutOutlets int
+	for lei, le := range in {
+
+		outlet, minLevel, err := parseOutlet(le)
+		if err != nil {
+			return nil, errors.Wrapf(err, "cannot parse outlet #%d", lei)
+		}
+		var _ logger.Outlet = WriterOutlet{}
+		var _ logger.Outlet = &SyslogOutlet{}
+		switch outlet.(type) {
+		case *SyslogOutlet:
+			syslogOutlets++
+		case WriterOutlet:
+			stdoutOutlets++
+		}
+
+		outlets.Add(outlet, minLevel)
+
+	}
+
+	if syslogOutlets > 1 {
+		return nil, errors.Errorf("can only define one 'syslog' outlet")
+	}
+	if stdoutOutlets > 1 {
+		return nil, errors.Errorf("can only define one 'stdout' outlet")
+	}
+
+	return outlets, nil
+
+}
+
+const (
+	SubsysReplication = "repl"
+	SubsysStreamrpc   = "rpc"
+	SubsyEndpoint     = "endpoint"
+)
+
+func WithSubsystemLoggers(ctx context.Context, log logger.Logger) context.Context {
+	ctx = replication.WithLogger(ctx, log.WithField(SubsysField, "repl"))
+	ctx = streamrpc.ContextWithLogger(ctx, streamrpcLogAdaptor{log.WithField(SubsysField, "rpc")})
+	ctx = endpoint.WithLogger(ctx, log.WithField(SubsysField, "endpoint"))
+	ctx = pruner.WithLogger(ctx, log.WithField(SubsysField, "pruning"))
+	ctx = snapper.WithLogger(ctx, log.WithField(SubsysField, "snapshot"))
+	ctx = serve.WithLogger(ctx, log.WithField(SubsysField, "serve"))
+	return ctx
+}
+
+func parseLogFormat(i interface{}) (f EntryFormatter, err error) {
+	var is string
+	switch j := i.(type) {
+	case string:
+		is = j
+	default:
+		return nil, errors.Errorf("invalid log format: wrong type: %T", i)
+	}
+
+	switch is {
+	case "human":
+		return &HumanFormatter{}, nil
+	case "logfmt":
+		return &LogfmtFormatter{}, nil
+	case "json":
+		return &JSONFormatter{}, nil
+	default:
+		return nil, errors.Errorf("invalid log format: '%s'", is)
+	}
+
+}
+
+func parseOutlet(in config.LoggingOutletEnum) (o logger.Outlet, level logger.Level, err error) {
+
+	parseCommon := func(common config.LoggingOutletCommon) (logger.Level, EntryFormatter, error) {
+		if common.Level == "" || common.Format == "" {
+			return 0, nil, errors.Errorf("must specify 'level' and 'format' field")
+		}
+
+		minLevel, err := logger.ParseLevel(common.Level)
+		if err != nil {
+			return 0, nil, errors.Wrap(err, "cannot parse 'level' field")
+		}
+		formatter, err := parseLogFormat(common.Format)
+		if err != nil {
+			return 0, nil, errors.Wrap(err, "cannot parse 'formatter' field")
+		}
+		return minLevel, formatter, nil
+	}
+
+	var f EntryFormatter
+
+	switch v := in.Ret.(type) {
+	case *config.StdoutLoggingOutlet:
+		level, f, err = parseCommon(v.LoggingOutletCommon)
+		if err != nil {
+			break
+		}
+		o, err = parseStdoutOutlet(v, f)
+	case *config.TCPLoggingOutlet:
+		level, f, err = parseCommon(v.LoggingOutletCommon)
+		if err != nil {
+			break
+		}
+		o, err = parseTCPOutlet(v, f)
+	case *config.SyslogLoggingOutlet:
+		level, f, err = parseCommon(v.LoggingOutletCommon)
+		if err != nil {
+			break
+		}
+		o, err = parseSyslogOutlet(v, f)
+	default:
+		panic(v)
+	}
+	return o, level, err
+}
+
+func parseStdoutOutlet(in *config.StdoutLoggingOutlet, formatter EntryFormatter) (WriterOutlet, error) {
+	flags := MetadataAll
+	writer := os.Stdout
+	if !isatty.IsTerminal(writer.Fd()) && !in.Time {
+		flags &= ^MetadataTime
+	}
+	if isatty.IsTerminal(writer.Fd()) && !in.Color {
+		flags &= ^MetadataColor
+	}
+
+	formatter.SetMetadataFlags(flags)
+	return WriterOutlet{
+		formatter,
+		os.Stdout,
+	}, nil
+}
+
+func parseTCPOutlet(in *config.TCPLoggingOutlet, formatter EntryFormatter) (out *TCPOutlet, err error) {
+	var tlsConfig *tls.Config
+	if in.TLS != nil {
+		tlsConfig, err = func(m *config.TCPLoggingOutletTLS, host string) (*tls.Config, error) {
+			clientCert, err := tls.LoadX509KeyPair(m.Cert, m.Key)
+			if err != nil {
+				return nil, errors.Wrap(err, "cannot load client cert")
+			}
+
+			var rootCAs *x509.CertPool
+			if m.CA == "" {
+				if rootCAs, err = x509.SystemCertPool(); err != nil {
+					return nil, errors.Wrap(err, "cannot open system cert pool")
+				}
+			} else {
+				rootCAs, err = tlsconf.ParseCAFile(m.CA)
+				if err != nil {
+					return nil, errors.Wrap(err, "cannot parse CA cert")
+				}
+			}
+			if rootCAs == nil {
+				panic("invariant violated")
+			}
+
+			return tlsconf.ClientAuthClient(host, rootCAs, clientCert)
+		}(in.TLS, in.Address)
+		if err != nil {
+			return nil, errors.New("cannot not parse TLS config in field 'tls'")
+		}
+	}
+
+	formatter.SetMetadataFlags(MetadataAll)
+	return NewTCPOutlet(formatter, in.Net, in.Address, tlsConfig, in.RetryInterval), nil
+
+}
+
+func parseSyslogOutlet(in *config.SyslogLoggingOutlet, formatter EntryFormatter) (out *SyslogOutlet, err error) {
+	out = &SyslogOutlet{}
+	out.Formatter = formatter
+	out.Formatter.SetMetadataFlags(MetadataNone)
+	out.RetryInterval = in.RetryInterval
+	return out, nil
+}

daemon/logging/logging_formatters.go

@@ -0,0 +1,207 @@
+package logging
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"github.com/fatih/color"
+	"github.com/go-logfmt/logfmt"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/logger"
+	"time"
+)
+
+const (
+	FieldLevel   = "level"
+	FieldMessage = "msg"
+	FieldTime    = "time"
+)
+
+const (
+	JobField    string = "job"
+	SubsysField string = "subsystem"
+)
+
+type MetadataFlags int64
+
+const (
+	MetadataTime MetadataFlags = 1 << iota
+	MetadataLevel
+	MetadataColor
+
+	MetadataNone MetadataFlags = 0
+	MetadataAll  MetadataFlags = ^0
+)
+
+type NoFormatter struct{}
+
+func (f NoFormatter) SetMetadataFlags(flags MetadataFlags) {}
+
+func (f NoFormatter) Format(e *logger.Entry) ([]byte, error) {
+	return []byte(e.Message), nil
+}
+
+type HumanFormatter struct {
+	metadataFlags MetadataFlags
+	ignoreFields  map[string]bool
+}
+
+const HumanFormatterDateFormat = time.RFC3339
+
+func (f *HumanFormatter) SetMetadataFlags(flags MetadataFlags) {
+	f.metadataFlags = flags
+}
+
+func (f *HumanFormatter) SetIgnoreFields(ignore []string) {
+	if ignore == nil {
+		f.ignoreFields = nil
+		return
+	}
+	f.ignoreFields = make(map[string]bool, len(ignore))
+
+	for _, field := range ignore {
+		f.ignoreFields[field] = true
+	}
+}
+
+func (f *HumanFormatter) ignored(field string) bool {
+	return f.ignoreFields != nil && f.ignoreFields[field]
+}
+
+func (f *HumanFormatter) Format(e *logger.Entry) (out []byte, err error) {
+
+	var line bytes.Buffer
+	col := color.New()
+	if f.metadataFlags&MetadataColor != 0 {
+		col = e.Color()
+	}
+
+	if f.metadataFlags&MetadataTime != 0 {
+		fmt.Fprintf(&line, "%s ", e.Time.Format(HumanFormatterDateFormat))
+	}
+	if f.metadataFlags&MetadataLevel != 0 {
+		fmt.Fprintf(&line, "[%s]", col.Sprint(e.Level.Short()))
+	}
+
+	prefixFields := []string{JobField, SubsysField}
+	prefixed := make(map[string]bool, len(prefixFields)+2)
+	for _, field := range prefixFields {
+		val, ok := e.Fields[field].(string)
+		if !ok {
+			continue
+		}
+		if !f.ignored(field) {
+			fmt.Fprintf(&line, "[%s]", col.Sprint(val))
+			prefixed[field] = true
+		}
+	}
+
+	if line.Len() > 0 {
+		fmt.Fprint(&line, ": ")
+	}
+	fmt.Fprint(&line, e.Message)
+
+	if len(e.Fields)-len(prefixed) > 0 {
+		for field, value := range e.Fields {
+			if prefixed[field] || f.ignored(field) {
+				continue
+			}
+			fmt.Fprintf(&line, " %s=%q", col.Sprint(field), fmt.Sprint(value))
+		}
+	}
+
+	return line.Bytes(), nil
+}
+
+type JSONFormatter struct {
+	metadataFlags MetadataFlags
+}
+
+func (f *JSONFormatter) SetMetadataFlags(flags MetadataFlags) {
+	f.metadataFlags = flags
+}
+
+func (f *JSONFormatter) Format(e *logger.Entry) ([]byte, error) {
+	data := make(logger.Fields, len(e.Fields)+3)
+	for k, v := range e.Fields {
+		switch v := v.(type) {
+		case error:
+			// Otherwise errors are ignored by `encoding/json`
+			// https://github.com/sirupsen/logrus/issues/137
+			data[k] = v.Error()
+		default:
+			_, err := json.Marshal(v)
+			if err != nil {
+				return nil, errors.Errorf("field is not JSON encodable: %s", k)
+			}
+			data[k] = v
+		}
+	}
+
+	data[FieldMessage] = e.Message
+	data[FieldTime] = e.Time.Format(time.RFC3339)
+	data[FieldLevel] = e.Level
+
+	return json.Marshal(data)
+
+}
+
+type LogfmtFormatter struct {
+	metadataFlags MetadataFlags
+}
+
+func (f *LogfmtFormatter) SetMetadataFlags(flags MetadataFlags) {
+	f.metadataFlags = flags
+}
+
+func (f *LogfmtFormatter) Format(e *logger.Entry) ([]byte, error) {
+	var buf bytes.Buffer
+	enc := logfmt.NewEncoder(&buf)
+
+	if f.metadataFlags&MetadataTime != 0 {
+		enc.EncodeKeyval(FieldTime, e.Time)
+	}
+	if f.metadataFlags&MetadataLevel != 0 {
+		enc.EncodeKeyval(FieldLevel, e.Level)
+	}
+
+	// at least try and put job and task in front
+	prefixed := make(map[string]bool, 2)
+	prefix := []string{JobField, SubsysField}
+	for _, pf := range prefix {
+		v, ok := e.Fields[pf]
+		if !ok {
+			break
+		}
+		if err := logfmtTryEncodeKeyval(enc, pf, v); err != nil {
+			return nil, err // unlikely
+		}
+		prefixed[pf] = true
+	}
+
+	enc.EncodeKeyval(FieldMessage, e.Message)
+
+	for k, v := range e.Fields {
+		if !prefixed[k] {
+			if err := logfmtTryEncodeKeyval(enc, k, v); err != nil {
+				return nil, err
+			}
+		}
+	}
+
+	return buf.Bytes(), nil
+}
+
+func logfmtTryEncodeKeyval(enc *logfmt.Encoder, field, value interface{}) error {
+
+	err := enc.EncodeKeyval(field, value)
+	switch err {
+	case nil: // ok
+		return nil
+	case logfmt.ErrUnsupportedValueType:
+		enc.EncodeKeyval(field, fmt.Sprintf("<%T>", value))
+		return nil
+	}
+	return errors.Wrapf(err, "cannot encode field '%s'", field)
+
+}

daemon/logging/logging_outlets.go

@@ -0,0 +1,166 @@
+package logging
+
+import (
+	"bytes"
+	"context"
+	"crypto/tls"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/logger"
+	"io"
+	"log/syslog"
+	"net"
+	"time"
+)
+
+type EntryFormatter interface {
+	SetMetadataFlags(flags MetadataFlags)
+	Format(e *logger.Entry) ([]byte, error)
+}
+
+type WriterOutlet struct {
+	formatter EntryFormatter
+	writer    io.Writer
+}
+
+func (h WriterOutlet) WriteEntry(entry logger.Entry) error {
+	bytes, err := h.formatter.Format(&entry)
+	if err != nil {
+		return err
+	}
+	_, err = h.writer.Write(bytes)
+	h.writer.Write([]byte("\n"))
+	return err
+}
+
+type TCPOutlet struct {
+	formatter EntryFormatter
+	// Specifies how much time must pass between a connection error and a reconnection attempt
+	// Log entries written to the outlet during this time interval are silently dropped.
+	connect   func(ctx context.Context) (net.Conn, error)
+	entryChan chan *bytes.Buffer
+}
+
+func NewTCPOutlet(formatter EntryFormatter, network, address string, tlsConfig *tls.Config, retryInterval time.Duration) *TCPOutlet {
+
+	connect := func(ctx context.Context) (conn net.Conn, err error) {
+		deadl, ok := ctx.Deadline()
+		if !ok {
+			deadl = time.Time{}
+		}
+		dialer := net.Dialer{
+			Deadline: deadl,
+		}
+		if tlsConfig != nil {
+			conn, err = tls.DialWithDialer(&dialer, network, address, tlsConfig)
+		} else {
+			conn, err = dialer.DialContext(ctx, network, address)
+		}
+		return
+	}
+
+	entryChan := make(chan *bytes.Buffer, 1) // allow one message in flight while previos is in io.Copy()
+
+	o := &TCPOutlet{
+		formatter: formatter,
+		connect:   connect,
+		entryChan: entryChan,
+	}
+
+	go o.outLoop(retryInterval)
+
+	return o
+}
+
+// FIXME: use this method
+func (h *TCPOutlet) Close() {
+	close(h.entryChan)
+}
+
+func (h *TCPOutlet) outLoop(retryInterval time.Duration) {
+
+	var retry time.Time
+	var conn net.Conn
+	for msg := range h.entryChan {
+		var err error
+		for conn == nil {
+			time.Sleep(time.Until(retry))
+			ctx, cancel := context.WithDeadline(context.TODO(), time.Now().Add(retryInterval))
+			conn, err = h.connect(ctx)
+			cancel()
+			if err != nil {
+				retry = time.Now().Add(retryInterval)
+				conn = nil
+			}
+		}
+		conn.SetWriteDeadline(time.Now().Add(retryInterval))
+		_, err = io.Copy(conn, msg)
+		if err != nil {
+			retry = time.Now().Add(retryInterval)
+			conn.Close()
+			conn = nil
+		}
+	}
+}
+
+func (h *TCPOutlet) WriteEntry(e logger.Entry) error {
+
+	ebytes, err := h.formatter.Format(&e)
+	if err != nil {
+		return err
+	}
+
+	buf := new(bytes.Buffer)
+	buf.Write(ebytes)
+	buf.WriteString("\n")
+
+	select {
+	case h.entryChan <- buf:
+		return nil
+	default:
+		return errors.New("connection broken or not fast enough")
+	}
+}
+
+type SyslogOutlet struct {
+	Formatter          EntryFormatter
+	RetryInterval      time.Duration
+	writer             *syslog.Writer
+	lastConnectAttempt time.Time
+}
+
+func (o *SyslogOutlet) WriteEntry(entry logger.Entry) error {
+
+	bytes, err := o.Formatter.Format(&entry)
+	if err != nil {
+		return err
+	}
+
+	s := string(bytes)
+
+	if o.writer == nil {
+		now := time.Now()
+		if now.Sub(o.lastConnectAttempt) < o.RetryInterval {
+			return nil // not an error toward logger
+		}
+		o.writer, err = syslog.New(syslog.LOG_LOCAL0, "zrepl")
+		o.lastConnectAttempt = time.Now()
+		if err != nil {
+			o.writer = nil
+			return err
+		}
+	}
+
+	switch entry.Level {
+	case logger.Debug:
+		return o.writer.Debug(s)
+	case logger.Info:
+		return o.writer.Info(s)
+	case logger.Warn:
+		return o.writer.Warning(s)
+	case logger.Error:
+		return o.writer.Err(s)
+	default:
+		return o.writer.Err(s) // write as error as reaching this case is in fact an error
+	}
+
+}

daemon/main.go

@@ -0,0 +1,16 @@
+package daemon
+
+import (
+	"github.com/zrepl/zrepl/cli"
+	"github.com/zrepl/zrepl/logger"
+)
+
+type Logger = logger.Logger
+
+var DaemonCmd = &cli.Subcommand {
+	Use:   "daemon",
+	Short: "run the zrepl daemon",
+	Run: func(subcommand *cli.Subcommand, args []string) error {
+		return Run(subcommand.Config())
+	},
+}

daemon/nethelpers/helpers.go

@@ -0,0 +1,49 @@
+package nethelpers
+
+import (
+	"github.com/pkg/errors"
+	"net"
+	"os"
+	"path/filepath"
+)
+
+func PreparePrivateSockpath(sockpath string) error {
+	sockdir := filepath.Dir(sockpath)
+	sdstat, err := os.Stat(sockdir)
+	if err != nil {
+		return errors.Wrapf(err, "cannot stat(2) '%s'", sockdir)
+	}
+	if !sdstat.IsDir() {
+		return errors.Errorf("not a directory: %s", sockdir)
+	}
+	p := sdstat.Mode().Perm()
+	if p&0007 != 0 {
+		return errors.Errorf("socket directory must not be world-accessible: %s (permissions are %#o)", sockdir, p)
+	}
+
+	// Maybe things have not been cleaned up before
+	s, err := os.Stat(sockpath)
+	if os.IsNotExist(err) {
+		return nil
+	}
+	if err != nil {
+		return errors.Wrapf(err, "cannot stat(2) '%s'", sockpath)
+	}
+	if s.Mode()&os.ModeSocket == 0 {
+		return errors.Errorf("unexpected file type at path '%s'", sockpath)
+	}
+	err = os.Remove(sockpath)
+	if err != nil {
+		return errors.Wrapf(err, "cannot remove presumably stale socket '%s'", sockpath)
+	}
+	return nil
+}
+
+func ListenUnixPrivate(sockaddr *net.UnixAddr) (*net.UnixListener, error) {
+
+	if err := PreparePrivateSockpath(sockaddr.Name); err != nil {
+		return nil, err
+	}
+
+	return net.ListenUnix("unix", sockaddr)
+}

daemon/pprof.go

@@ -0,0 +1,80 @@
+package daemon
+
+import (
+	"net/http"
+	// FIXME: importing this package has the side-effect of poisoning the http.DefaultServeMux
+	// FIXME: with the /debug/pprof endpoints
+	"context"
+	"net"
+	"net/http/pprof"
+)
+
+type pprofServer struct {
+	cc       chan PprofServerControlMsg
+	state    PprofServerControlMsg
+	listener net.Listener
+}
+
+type PprofServerControlMsg struct {
+	// Whether the server should listen for requests on the given address
+	Run bool
+	// Must be set if Run is true, undefined otherwise
+	HttpListenAddress string
+}
+
+func NewPProfServer(ctx context.Context) *pprofServer {
+
+	s := &pprofServer{
+		cc: make(chan PprofServerControlMsg),
+	}
+
+	go s.controlLoop(ctx)
+	return s
+}
+
+func (s *pprofServer) controlLoop(ctx context.Context) {
+outer:
+	for {
+
+		var msg PprofServerControlMsg
+		select {
+		case <-ctx.Done():
+			if s.listener != nil {
+				s.listener.Close()
+			}
+			break outer
+		case msg = <-s.cc:
+			// proceed
+		}
+
+		var err error
+		if msg.Run && s.listener == nil {
+
+			s.listener, err = net.Listen("tcp", msg.HttpListenAddress)
+			if err != nil {
+				s.listener = nil
+				continue
+			}
+
+			// FIXME: because net/http/pprof does not provide a mux,
+			mux := http.NewServeMux()
+			mux.Handle("/debug/pprof/", http.HandlerFunc(pprof.Index))
+			mux.Handle("/debug/pprof/cmdline", http.HandlerFunc(pprof.Cmdline))
+			mux.Handle("/debug/pprof/profile", http.HandlerFunc(pprof.Profile))
+			mux.Handle("/debug/pprof/symbol", http.HandlerFunc(pprof.Symbol))
+			mux.Handle("/debug/pprof/trace", http.HandlerFunc(pprof.Trace))
+			go http.Serve(s.listener, mux)
+			continue
+		}
+
+		if !msg.Run && s.listener != nil {
+			s.listener.Close()
+			s.listener = nil
+			continue
+		}
+	}
+}
+
+func (s *pprofServer) Control(msg PprofServerControlMsg) {
+	s.cc <- msg
+}

daemon/prometheus.go

@@ -0,0 +1,89 @@
+package daemon
+
+import (
+	"context"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promhttp"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/job"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/zfs"
+	"net"
+	"net/http"
+)
+
+type prometheusJob struct {
+	listen string
+}
+
+func newPrometheusJobFromConfig(in *config.PrometheusMonitoring) (*prometheusJob, error) {
+	if _, _, err := net.SplitHostPort(in.Listen); err != nil {
+		return nil, err
+	}
+	return &prometheusJob{in.Listen}, nil
+}
+
+var prom struct {
+	taskLogEntries         *prometheus.CounterVec
+}
+
+func init() {
+	prom.taskLogEntries = prometheus.NewCounterVec(prometheus.CounterOpts{
+		Namespace: "zrepl",
+		Subsystem: "daemon",
+		Name:      "log_entries",
+		Help:      "number of log entries per job task and level",
+	}, []string{"zrepl_job", "level"})
+	prometheus.MustRegister(prom.taskLogEntries)
+}
+
+func (j *prometheusJob) Name() string { return jobNamePrometheus }
+
+func (j *prometheusJob) Status() *job.Status { return &job.Status{Type: job.TypeInternal} }
+
+func (j *prometheusJob) RegisterMetrics(registerer prometheus.Registerer) {}
+
+func (j *prometheusJob) Run(ctx context.Context) {
+
+	if err := zfs.PrometheusRegister(prometheus.DefaultRegisterer); err != nil {
+		panic(err)
+	}
+
+	log := job.GetLogger(ctx)
+
+	l, err := net.Listen("tcp", j.listen)
+	if err != nil {
+		log.WithError(err).Error("cannot listen")
+	}
+	go func() {
+		select {
+		case <-ctx.Done():
+			l.Close()
+		}
+	}()
+
+	mux := http.NewServeMux()
+	mux.Handle("/metrics", promhttp.Handler())
+
+	err = http.Serve(l, mux)
+	if err != nil {
+		log.WithError(err).Error("error while serving")
+	}
+
+}
+
+type prometheusJobOutlet struct {
+	jobName string
+}
+
+var _ logger.Outlet = prometheusJobOutlet{}
+
+func newPrometheusLogOutlet(jobName string) prometheusJobOutlet {
+	return prometheusJobOutlet{jobName}
+}
+
+func (o prometheusJobOutlet) WriteEntry(entry logger.Entry) error {
+	prom.taskLogEntries.WithLabelValues(o.jobName, entry.Level.String()).Inc()
+	return nil
+}
+

daemon/pruner/pruner.go

@@ -0,0 +1,584 @@
+package pruner
+
+import (
+	"context"
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/pruning"
+	"github.com/zrepl/zrepl/replication/pdu"
+	"github.com/zrepl/zrepl/util/envconst"
+	"github.com/zrepl/zrepl/util/watchdog"
+	"github.com/problame/go-streamrpc"
+	"net"
+	"sort"
+	"strings"
+	"sync"
+	"time"
+)
+
+// Try to keep it compatible with gitub.com/zrepl/zrepl/replication.Endpoint
+type History interface {
+	ReplicationCursor(ctx context.Context, req *pdu.ReplicationCursorReq) (*pdu.ReplicationCursorRes, error)
+}
+
+type Target interface {
+	ListFilesystems(ctx context.Context) ([]*pdu.Filesystem, error)
+	ListFilesystemVersions(ctx context.Context, fs string) ([]*pdu.FilesystemVersion, error) // fix depS
+	DestroySnapshots(ctx context.Context, req *pdu.DestroySnapshotsReq) (*pdu.DestroySnapshotsRes, error)
+}
+
+type Logger = logger.Logger
+
+type contextKey int
+
+const contextKeyLogger contextKey = 0
+
+func WithLogger(ctx context.Context, log Logger) context.Context {
+	return context.WithValue(ctx, contextKeyLogger, log)
+}
+
+func GetLogger(ctx context.Context) Logger {
+	if l, ok := ctx.Value(contextKeyLogger).(Logger); ok {
+		return l
+	}
+	return logger.NewNullLogger()
+}
+
+type args struct {
+	ctx                            context.Context
+	target                         Target
+	receiver                       History
+	rules                          []pruning.KeepRule
+	retryWait                      time.Duration
+	considerSnapAtCursorReplicated bool
+	promPruneSecs prometheus.Observer
+}
+
+type Pruner struct {
+	args args
+
+	Progress watchdog.KeepAlive
+
+	mtx sync.RWMutex
+
+	state State
+
+	// State ErrWait|ErrPerm
+	sleepUntil time.Time
+	err        error
+
+	// State Exec
+	execQueue *execQueue
+}
+
+type PrunerFactory struct {
+	senderRules                    []pruning.KeepRule
+	receiverRules                  []pruning.KeepRule
+	retryWait                      time.Duration
+	considerSnapAtCursorReplicated bool
+	promPruneSecs *prometheus.HistogramVec
+}
+
+func checkContainsKeep1(rules []pruning.KeepRule) error {
+	if len(rules) == 0 {
+		return nil //No keep rules means keep all - ok
+	}
+	for _, e := range rules {
+		switch e.(type) {
+		case *pruning.KeepLastN:
+			return nil
+		}
+	}
+	return errors.New("sender keep rules must contain last_n or be empty so that the last snapshot is definitely kept")
+}
+
+func NewPrunerFactory(in config.PruningSenderReceiver, promPruneSecs *prometheus.HistogramVec) (*PrunerFactory, error) {
+	keepRulesReceiver, err := pruning.RulesFromConfig(in.KeepReceiver)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot build receiver pruning rules")
+	}
+
+	keepRulesSender, err := pruning.RulesFromConfig(in.KeepSender)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot build sender pruning rules")
+	}
+
+	considerSnapAtCursorReplicated := false
+	for _, r := range in.KeepSender {
+		knr, ok := r.Ret.(*config.PruneKeepNotReplicated)
+		if !ok {
+			continue
+		}
+		considerSnapAtCursorReplicated = considerSnapAtCursorReplicated || !knr.KeepSnapshotAtCursor
+	}
+	f := &PrunerFactory{
+		senderRules: keepRulesSender,
+		receiverRules: keepRulesReceiver,
+		retryWait: envconst.Duration("ZREPL_PRUNER_RETRY_INTERVAL", 10 * time.Second),
+		considerSnapAtCursorReplicated: considerSnapAtCursorReplicated,
+		promPruneSecs: promPruneSecs,
+	}
+	return f, nil
+}
+
+func (f *PrunerFactory) BuildSenderPruner(ctx context.Context, target Target, receiver History) *Pruner {
+	p := &Pruner{
+		args: args{
+			WithLogger(ctx, GetLogger(ctx).WithField("prune_side", "sender")),
+			target,
+			receiver,
+			f.senderRules,
+			f.retryWait,
+			f.considerSnapAtCursorReplicated,
+			f.promPruneSecs.WithLabelValues("sender"),
+		},
+		state: Plan,
+	}
+	return p
+}
+
+func (f *PrunerFactory) BuildReceiverPruner(ctx context.Context, target Target, receiver History) *Pruner {
+	p := &Pruner{
+		args: args{
+			WithLogger(ctx, GetLogger(ctx).WithField("prune_side", "receiver")),
+			target,
+			receiver,
+			f.receiverRules,
+			f.retryWait,
+			false, // senseless here anyways
+			f.promPruneSecs.WithLabelValues("receiver"),
+		},
+		state: Plan,
+	}
+	return p
+}
+
+//go:generate enumer -type=State
+type State int
+
+const (
+	Plan State = 1 << iota
+	PlanWait
+	Exec
+	ExecWait
+	ErrPerm
+	Done
+)
+
+func (s State) statefunc() state {
+	var statemap = map[State]state{
+		Plan:     statePlan,
+		PlanWait: statePlanWait,
+		Exec:     stateExec,
+		ExecWait: stateExecWait,
+		ErrPerm:  nil,
+		Done:     nil,
+	}
+	return statemap[s]
+}
+
+func (s State) IsTerminal() bool {
+	return s.statefunc() == nil
+}
+
+type updater func(func(*Pruner)) State
+type state func(args *args, u updater) state
+
+func (p *Pruner) Prune() {
+	p.prune(p.args)
+}
+
+func (p *Pruner) prune(args args) {
+	s := p.state.statefunc()
+	for s != nil {
+		pre := p.state
+		s = s(&args, func(f func(*Pruner)) State {
+			p.mtx.Lock()
+			defer p.mtx.Unlock()
+			f(p)
+			return p.state
+		})
+		post := p.state
+		GetLogger(args.ctx).
+			WithField("transition", fmt.Sprintf("%s=>%s", pre, post)).
+			Debug("state transition")
+		if err := p.Error(); err != nil {
+			GetLogger(args.ctx).
+				WithError(p.err).
+				WithField("state", post.String()).
+				Error("entering error state after error")
+		}
+	}
+}
+
+type Report struct {
+	State string
+	SleepUntil time.Time
+	Error string
+	Pending, Completed []FSReport
+}
+
+type FSReport struct {
+	Filesystem string
+	SnapshotList, DestroyList []SnapshotReport
+	ErrorCount int
+	LastError string
+}
+
+type SnapshotReport struct {
+	Name string
+	Replicated bool
+	Date time.Time
+}
+
+func (p *Pruner) Report() *Report {
+	p.mtx.Lock()
+	defer p.mtx.Unlock()
+
+	r := Report{State: p.state.String()}
+
+	if p.state & (PlanWait|ExecWait) != 0 {
+		r.SleepUntil = p.sleepUntil
+	}
+	if p.state & (PlanWait|ExecWait|ErrPerm) != 0 {
+		if p.err != nil {
+			r.Error = p.err.Error()
+		}
+	}
+
+	if p.execQueue != nil {
+		r.Pending, r.Completed = p.execQueue.Report()
+	}
+
+	return &r
+}
+
+func (p *Pruner) State() State {
+	p.mtx.Lock()
+	defer p.mtx.Unlock()
+	return p.state
+}
+
+func (p *Pruner) Error() error {
+	p.mtx.Lock()
+	defer p.mtx.Unlock()
+	if p.state & (PlanWait|ExecWait|ErrPerm) != 0 {
+		return p.err
+	}
+	return nil
+}
+
+type fs struct {
+	path  string
+
+	// snapshots presented by target
+	// (type snapshot)
+	snaps []pruning.Snapshot
+	// destroy list returned by pruning.PruneSnapshots(snaps)
+	// (type snapshot)
+	destroyList []pruning.Snapshot
+
+	mtx sync.RWMutex
+
+	// only during Exec state, also used by execQueue
+	execErrLast error
+	execErrCount int
+
+}
+
+func (f *fs) Report() FSReport {
+	f.mtx.Lock()
+	defer f.mtx.Unlock()
+
+	r := FSReport{}
+	r.Filesystem = f.path
+	r.ErrorCount = f.execErrCount
+	if f.execErrLast != nil {
+		r.LastError = f.execErrLast.Error()
+	}
+
+	r.SnapshotList = make([]SnapshotReport, len(f.snaps))
+	for i, snap := range f.snaps {
+		r.SnapshotList[i] = snap.(snapshot).Report()
+	}
+
+	r.DestroyList = make([]SnapshotReport, len(f.destroyList))
+	for i, snap := range f.destroyList{
+		r.DestroyList[i] = snap.(snapshot).Report()
+	}
+
+	return r
+}
+
+type snapshot struct {
+	replicated bool
+	date       time.Time
+	fsv        *pdu.FilesystemVersion
+}
+
+func (s snapshot) Report() SnapshotReport {
+	return SnapshotReport{
+		Name:       s.Name(),
+		Replicated: s.Replicated(),
+		Date:       s.Date(),
+	}
+}
+
+var _ pruning.Snapshot = snapshot{}
+
+func (s snapshot) Name() string { return s.fsv.Name }
+
+func (s snapshot) Replicated() bool { return s.replicated }
+
+func (s snapshot) Date() time.Time { return s.date }
+
+type Error interface {
+	error
+	Temporary() bool
+}
+
+var _ Error = net.Error(nil)
+var _ Error = streamrpc.Error(nil)
+
+func shouldRetry(e error) bool {
+	if neterr, ok := e.(net.Error); ok {
+		return neterr.Temporary()
+	}
+	return false
+}
+
+func onErr(u updater, e error) state {
+	return u(func(p *Pruner) {
+		p.err = e
+		if !shouldRetry(e) {
+			p.state = ErrPerm
+			return
+		}
+		switch p.state {
+		case Plan:
+			p.state = PlanWait
+		case Exec:
+			p.state = ExecWait
+		default:
+			panic(p.state)
+		}
+	}).statefunc()
+}
+
+func statePlan(a *args, u updater) state {
+
+	ctx, target, receiver := a.ctx, a.target, a.receiver
+	var ka *watchdog.KeepAlive
+	u(func(pruner *Pruner) {
+		ka = &pruner.Progress
+	})
+
+	tfss, err := target.ListFilesystems(ctx)
+	if err != nil {
+		return onErr(u, err)
+	}
+
+	pfss := make([]*fs, len(tfss))
+	for i, tfs := range tfss {
+
+		l := GetLogger(ctx).WithField("fs", tfs.Path)
+		l.Debug("plan filesystem")
+
+
+		pfs := &fs{
+			path:  tfs.Path,
+		}
+		pfss[i] = pfs
+
+		tfsvs, err := target.ListFilesystemVersions(ctx, tfs.Path)
+		if err != nil {
+			l.WithError(err).Error("cannot list filesystem versions")
+			return onErr(u, err)
+		}
+		// no progress here since we could run in a live-lock (must have used target AND receiver before progress)
+
+		pfs.snaps = make([]pruning.Snapshot, 0, len(tfsvs))
+
+		rcReq := &pdu.ReplicationCursorReq{
+			Filesystem: tfs.Path,
+			Op:         &pdu.ReplicationCursorReq_Get{
+				Get: &pdu.ReplicationCursorReq_GetOp{},
+			},
+		}
+		rc, err := receiver.ReplicationCursor(ctx, rcReq)
+		if err != nil {
+			l.WithError(err).Error("cannot get replication cursor")
+			return onErr(u, err)
+		}
+		if rc.GetError() != "" {
+			l.WithField("reqErr", rc.GetError()).Error("cannot get replication cursor")
+			return onErr(u, err)
+		}
+		ka.MadeProgress()
+
+
+		// scan from older to newer, all snapshots older than cursor are interpreted as replicated
+		sort.Slice(tfsvs, func(i, j int) bool {
+			return tfsvs[i].CreateTXG < tfsvs[j].CreateTXG
+		})
+
+		haveCursorSnapshot := false
+		for _, tfsv := range tfsvs {
+			if tfsv.Type != pdu.FilesystemVersion_Snapshot {
+				continue
+			}
+			if tfsv.Guid == rc.GetGuid() {
+				haveCursorSnapshot = true
+			}
+		}
+		preCursor := haveCursorSnapshot
+		for _, tfsv := range tfsvs {
+			if tfsv.Type != pdu.FilesystemVersion_Snapshot {
+				continue
+			}
+			creation, err := tfsv.CreationAsTime()
+			if err != nil {
+				err := fmt.Errorf("%s%s has invalid creation date: %s", tfs, tfsv.RelName(), err)
+				l.WithError(err).
+					WithField("tfsv", tfsv.RelName()).
+					Error("error with fileesystem version")
+				return onErr(u, err)
+			}
+			// note that we cannot use CreateTXG because target and receiver could be on different pools
+			atCursor := tfsv.Guid == rc.GetGuid()
+			preCursor = preCursor && !atCursor
+			pfs.snaps = append(pfs.snaps, snapshot{
+				replicated: preCursor || (a.considerSnapAtCursorReplicated && atCursor),
+				date:       creation,
+				fsv:        tfsv,
+			})
+		}
+		if preCursor {
+			err := fmt.Errorf("replication cursor not found in prune target filesystem versions")
+			l.Error(err.Error())
+			return onErr(u, err)
+		}
+
+		// Apply prune rules
+		pfs.destroyList = pruning.PruneSnapshots(pfs.snaps, a.rules)
+		ka.MadeProgress()
+	}
+
+	return u(func(pruner *Pruner) {
+		pruner.Progress.MadeProgress()
+		pruner.execQueue = newExecQueue(len(pfss))
+		for _, pfs := range pfss {
+			pruner.execQueue.Put(pfs, nil, false)
+		}
+		pruner.state = Exec
+	}).statefunc()
+}
+
+func stateExec(a *args, u updater) state {
+
+	var pfs *fs
+	state := u(func(pruner *Pruner) {
+		pfs = pruner.execQueue.Pop()
+		if pfs == nil {
+			nextState := Done
+			if pruner.execQueue.HasCompletedFSWithErrors() {
+				nextState = ErrPerm
+			}
+			pruner.state = nextState
+			return
+		}
+	})
+	if state != Exec {
+		return state.statefunc()
+	}
+
+	destroyList := make([]*pdu.FilesystemVersion, len(pfs.destroyList))
+	for i := range destroyList {
+		destroyList[i] = pfs.destroyList[i].(snapshot).fsv
+		GetLogger(a.ctx).
+			WithField("fs", pfs.path).
+			WithField("destroy_snap", destroyList[i].Name).
+			Debug("policy destroys snapshot")
+	}
+	req := pdu.DestroySnapshotsReq{
+		Filesystem: pfs.path,
+		Snapshots:  destroyList,
+	}
+	GetLogger(a.ctx).WithField("fs", pfs.path).Debug("destroying snapshots")
+	res, err := a.target.DestroySnapshots(a.ctx, &req)
+	if err != nil {
+		u(func(pruner *Pruner) {
+			pruner.execQueue.Put(pfs, err, false)
+		})
+		return onErr(u, err)
+	}
+	// check if all snapshots were destroyed
+	destroyResults := make(map[string]*pdu.DestroySnapshotRes)
+	for _, fsres := range res.Results {
+		destroyResults[fsres.Snapshot.Name] = fsres
+	}
+	err = nil
+	destroyFails := make([]*pdu.DestroySnapshotRes, 0)
+	for _, reqDestroy := range destroyList {
+		 res, ok := destroyResults[reqDestroy.Name]
+		 if !ok {
+		 	err = fmt.Errorf("missing destroy-result for %s", reqDestroy.RelName())
+		 	break
+		 } else if res.Error != "" {
+		 	destroyFails = append(destroyFails, res)
+		 }
+	}
+	if err == nil && len(destroyFails) > 0 {
+		names := make([]string, len(destroyFails))
+		pairs := make([]string, len(destroyFails))
+		allSame := true
+		lastMsg := destroyFails[0].Error
+		for i := 0; i < len(destroyFails); i++{
+			allSame = allSame && destroyFails[i].Error == lastMsg
+			relname := destroyFails[i].Snapshot.RelName()
+			names[i] = relname
+			pairs[i] = fmt.Sprintf("(%s: %s)", relname, destroyFails[i].Error)
+		}
+		if allSame {
+			err = fmt.Errorf("destroys failed %s: %s",
+				strings.Join(names, ", "), lastMsg)
+		} else {
+			err = fmt.Errorf("destroys failed: %s", strings.Join(pairs, ", "))
+		}
+	}
+	u(func(pruner *Pruner) {
+		pruner.execQueue.Put(pfs, err, err == nil)
+	})
+	if err != nil {
+		GetLogger(a.ctx).WithError(err).Error("target could not destroy snapshots")
+		return onErr(u, err)
+	}
+
+	return u(func(pruner *Pruner) {
+		pruner.Progress.MadeProgress()
+	}).statefunc()
+}
+
+func stateExecWait(a *args, u updater) state {
+	return doWait(Exec, a, u)
+}
+
+func statePlanWait(a *args, u updater) state {
+	return doWait(Plan, a, u)
+}
+
+func doWait(goback State, a *args, u updater) state {
+	timer := time.NewTimer(a.retryWait)
+	defer timer.Stop()
+	select {
+	case <-timer.C:
+		return u(func(pruner *Pruner) {
+			pruner.state = goback
+		}).statefunc()
+	case <-a.ctx.Done():
+		return onErr(u, a.ctx.Err())
+	}
+}

daemon/pruner/pruner_queue.go

@@ -0,0 +1,89 @@
+package pruner
+
+import (
+	"sort"
+	"strings"
+	"sync"
+)
+
+type execQueue struct {
+	mtx sync.Mutex
+	pending, completed []*fs
+}
+
+func newExecQueue(cap int) *execQueue {
+	q := execQueue{
+		pending: make([]*fs, 0, cap),
+		completed: make([]*fs, 0, cap),
+	}
+	return &q
+}
+
+func (q *execQueue) Report() (pending, completed []FSReport) {
+	q.mtx.Lock()
+	defer q.mtx.Unlock()
+
+	pending = make([]FSReport, len(q.pending))
+	for i, fs := range q.pending {
+		pending[i] = fs.Report()
+	}
+	completed = make([]FSReport, len(q.completed))
+	for i, fs := range q.completed {
+		completed[i] = fs.Report()
+	}
+
+	return pending, completed
+}
+
+func (q *execQueue) HasCompletedFSWithErrors() bool {
+	q.mtx.Lock()
+	defer q.mtx.Unlock()
+	for _, fs := range q.completed {
+		if fs.execErrLast != nil {
+			return true
+		}
+	}
+	return false
+}
+
+func (q *execQueue) Pop() *fs {
+	if len(q.pending) == 0 {
+		return nil
+	}
+	fs := q.pending[0]
+	q.pending = q.pending[1:]
+	return fs
+}
+
+func(q *execQueue) Put(fs *fs, err error, done bool) {
+	fs.mtx.Lock()
+	fs.execErrLast = err
+	if err != nil {
+		fs.execErrCount++
+	}
+	if done || (err != nil && !shouldRetry(fs.execErrLast)) {
+		fs.mtx.Unlock()
+		q.mtx.Lock()
+		q.completed = append(q.completed, fs)
+		q.mtx.Unlock()
+		return
+	}
+	fs.mtx.Unlock()
+
+	q.mtx.Lock()
+	// inefficient priority q
+	q.pending = append(q.pending, fs)
+	sort.SliceStable(q.pending, func(i, j int) bool {
+		q.pending[i].mtx.Lock()
+		defer q.pending[i].mtx.Unlock()
+		q.pending[j].mtx.Lock()
+		defer q.pending[j].mtx.Unlock()
+		if q.pending[i].execErrCount != q.pending[j].execErrCount {
+			return q.pending[i].execErrCount < q.pending[j].execErrCount
+		}
+		return strings.Compare(q.pending[i].path, q.pending[j].path) == -1
+	})
+	q.mtx.Unlock()
+
+
+}

daemon/pruner/pruner_test.go

@@ -0,0 +1,205 @@
+package pruner
+
+import (
+	"context"
+	"fmt"
+	"github.com/stretchr/testify/assert"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/pruning"
+	"github.com/zrepl/zrepl/replication/pdu"
+	"net"
+	"testing"
+	"time"
+)
+
+type mockFS struct {
+	path  string
+	snaps []string
+}
+
+func (m *mockFS) Filesystem() *pdu.Filesystem {
+	return &pdu.Filesystem{
+		Path: m.path,
+	}
+}
+
+func (m *mockFS) FilesystemVersions() []*pdu.FilesystemVersion {
+	versions := make([]*pdu.FilesystemVersion, len(m.snaps))
+	for i, v := range m.snaps {
+		versions[i] = &pdu.FilesystemVersion{
+			Type:     pdu.FilesystemVersion_Snapshot,
+			Name:     v,
+			Creation: pdu.FilesystemVersionCreation(time.Unix(0, 0)),
+			Guid: uint64(i),
+		}
+	}
+	return versions
+}
+
+type mockTarget struct {
+	fss                []mockFS
+	destroyed          map[string][]string
+	listVersionsErrs   map[string][]error
+	listFilesystemsErr []error
+	destroyErrs        map[string][]error
+}
+
+func (t *mockTarget) ListFilesystems(ctx context.Context) ([]*pdu.Filesystem, error) {
+	if len(t.listFilesystemsErr) > 0 {
+		e := t.listFilesystemsErr[0]
+		t.listFilesystemsErr = t.listFilesystemsErr[1:]
+		return nil, e
+	}
+	fss := make([]*pdu.Filesystem, len(t.fss))
+	for i := range fss {
+		fss[i] = t.fss[i].Filesystem()
+	}
+	return fss, nil
+}
+
+func (t *mockTarget) ListFilesystemVersions(ctx context.Context, fs string) ([]*pdu.FilesystemVersion, error) {
+	if len(t.listVersionsErrs[fs]) != 0 {
+		e := t.listVersionsErrs[fs][0]
+		t.listVersionsErrs[fs] = t.listVersionsErrs[fs][1:]
+		return nil, e
+	}
+
+	for _, mfs := range t.fss {
+		if mfs.path != fs {
+			continue
+		}
+		return mfs.FilesystemVersions(), nil
+	}
+	return nil, fmt.Errorf("filesystem %s does not exist", fs)
+}
+
+func (t *mockTarget) DestroySnapshots(ctx context.Context, req *pdu.DestroySnapshotsReq) (*pdu.DestroySnapshotsRes, error) {
+	fs, snaps := req.Filesystem, req.Snapshots
+	if len(t.destroyErrs[fs]) != 0 {
+		e := t.destroyErrs[fs][0]
+		t.destroyErrs[fs] = t.destroyErrs[fs][1:]
+		return nil, e
+	}
+	destroyed := t.destroyed[fs]
+	res := make([]*pdu.DestroySnapshotRes, len(snaps))
+	for i, s := range snaps {
+		destroyed = append(destroyed, s.Name)
+		res[i] = &pdu.DestroySnapshotRes{Error: "", Snapshot: s}
+	}
+	t.destroyed[fs] = destroyed
+	return &pdu.DestroySnapshotsRes{Results: res}, nil
+}
+
+type mockCursor struct {
+	snapname string
+	guid uint64
+}
+type mockHistory struct {
+	errs map[string][]error
+	cursors map[string]*mockCursor
+}
+
+func (r *mockHistory) ReplicationCursor(ctx context.Context, req *pdu.ReplicationCursorReq) (*pdu.ReplicationCursorRes, error) {
+	fs := req.Filesystem
+	if len(r.errs[fs]) > 0 {
+		e := r.errs[fs][0]
+		r.errs[fs] = r.errs[fs][1:]
+		return nil, e
+	}
+	return &pdu.ReplicationCursorRes{Result: &pdu.ReplicationCursorRes_Guid{Guid: 0}}, nil
+}
+
+type stubNetErr struct {
+	msg string
+	temporary, timeout bool
+}
+
+var _ net.Error = stubNetErr{}
+
+func (e stubNetErr) Error() string {
+	return e.msg
+}
+
+func (e stubNetErr) Temporary() bool { return e.temporary }
+
+func (e stubNetErr) Timeout() bool { return e.timeout }
+
+func TestPruner_Prune(t *testing.T) {
+
+	var _ net.Error = &net.OpError{} // we use it below
+	target := &mockTarget{
+		listFilesystemsErr: []error{
+			stubNetErr{msg: "fakerror0", temporary: true},
+		},
+		listVersionsErrs: map[string][]error{
+			"zroot/foo": {
+				stubNetErr{msg: "fakeerror1", temporary: true},
+				stubNetErr{msg: "fakeerror2", temporary: true,},
+			},
+		},
+		destroyErrs: map[string][]error{
+			"zroot/baz": {
+				stubNetErr{msg: "fakeerror3", temporary: true}, // first error puts it back in the queue
+				stubNetErr{msg:"permanent error"}, // so it will be last when pruner gives up due to permanent err
+			},
+		},
+		destroyed: make(map[string][]string),
+		fss: []mockFS{
+			{
+				path: "zroot/foo",
+				snaps: []string{
+					"keep_a",
+					"keep_b",
+					"drop_c",
+					"keep_d",
+				},
+			},
+			{
+				path: "zroot/bar",
+				snaps: []string{
+					"keep_e",
+					"keep_f",
+					"drop_g",
+				},
+			},
+			{
+				path: "zroot/baz",
+				snaps: []string{
+					"keep_h",
+					"drop_i",
+				},
+			},
+		},
+	}
+	history := &mockHistory{
+		errs: map[string][]error{
+			"zroot/foo": {
+				stubNetErr{msg: "fakeerror4", temporary: true},
+			},
+		},
+	}
+
+	keepRules := []pruning.KeepRule{pruning.MustKeepRegex("^keep")}
+
+	p := Pruner{
+		args: args{
+			ctx: WithLogger(context.Background(), logger.NewTestLogger(t)),
+			target: target,
+			receiver: history,
+			rules: keepRules,
+			retryWait: 10*time.Millisecond,
+		},
+		state: Plan,
+	}
+	p.Prune()
+
+	exp := map[string][]string{
+		"zroot/foo": {"drop_c"},
+		"zroot/bar": {"drop_g"},
+	}
+
+	assert.Equal(t, exp, target.destroyed)
+
+	//assert.Equal(t, map[string][]error{}, target.listVersionsErrs, "retried")
+
+}

daemon/pruner/state_enumer.go

@@ -0,0 +1,76 @@
+// Code generated by "enumer -type=State"; DO NOT EDIT.
+
+package pruner
+
+import (
+	"fmt"
+)
+
+const (
+	_StateName_0 = "PlanPlanWait"
+	_StateName_1 = "Exec"
+	_StateName_2 = "ExecWait"
+	_StateName_3 = "ErrPerm"
+	_StateName_4 = "Done"
+)
+
+var (
+	_StateIndex_0 = [...]uint8{0, 4, 12}
+	_StateIndex_1 = [...]uint8{0, 4}
+	_StateIndex_2 = [...]uint8{0, 8}
+	_StateIndex_3 = [...]uint8{0, 7}
+	_StateIndex_4 = [...]uint8{0, 4}
+)
+
+func (i State) String() string {
+	switch {
+	case 1 <= i && i <= 2:
+		i -= 1
+		return _StateName_0[_StateIndex_0[i]:_StateIndex_0[i+1]]
+	case i == 4:
+		return _StateName_1
+	case i == 8:
+		return _StateName_2
+	case i == 16:
+		return _StateName_3
+	case i == 32:
+		return _StateName_4
+	default:
+		return fmt.Sprintf("State(%d)", i)
+	}
+}
+
+var _StateValues = []State{1, 2, 4, 8, 16, 32}
+
+var _StateNameToValueMap = map[string]State{
+	_StateName_0[0:4]:  1,
+	_StateName_0[4:12]: 2,
+	_StateName_1[0:4]:  4,
+	_StateName_2[0:8]:  8,
+	_StateName_3[0:7]:  16,
+	_StateName_4[0:4]:  32,
+}
+
+// StateString retrieves an enum value from the enum constants string name.
+// Throws an error if the param is not part of the enum.
+func StateString(s string) (State, error) {
+	if val, ok := _StateNameToValueMap[s]; ok {
+		return val, nil
+	}
+	return 0, fmt.Errorf("%s does not belong to State values", s)
+}
+
+// StateValues returns all values of the enum
+func StateValues() []State {
+	return _StateValues
+}
+
+// IsAState returns "true" if the value is listed in the enum definition. "false" otherwise
+func (i State) IsAState() bool {
+	for _, v := range _StateValues {
+		if i == v {
+			return true
+		}
+	}
+	return false
+}

daemon/snapper/snapper.go

@@ -0,0 +1,369 @@
+package snapper
+
+import (
+	"github.com/zrepl/zrepl/config"
+	"github.com/pkg/errors"
+	"time"
+	"context"
+	"github.com/zrepl/zrepl/daemon/filters"
+	"fmt"
+	"github.com/zrepl/zrepl/zfs"
+	"sort"
+	"github.com/zrepl/zrepl/logger"
+	"sync"
+)
+
+
+//go:generate stringer -type=SnapState
+type SnapState uint
+
+const (
+	SnapPending SnapState = 1 << iota
+	SnapStarted
+	SnapDone
+	SnapError
+)
+
+type snapProgress struct {
+	state SnapState
+
+	// SnapStarted, SnapDone, SnapError
+	name string
+	startAt time.Time
+
+	// SnapDone
+	doneAt time.Time
+
+	// SnapErr
+	err error
+}
+
+type args struct {
+	ctx            context.Context
+	log            Logger
+	prefix         string
+	interval       time.Duration
+	fsf            *filters.DatasetMapFilter
+	snapshotsTaken chan<-struct{}
+}
+
+type Snapper struct {
+	args args
+
+	mtx sync.Mutex
+	state State
+
+	// set in state Plan, used in Waiting
+	lastInvocation time.Time
+
+	// valid for state Snapshotting
+	plan map[*zfs.DatasetPath]snapProgress
+
+	// valid for state SyncUp and Waiting
+	sleepUntil time.Time
+
+	// valid for state Err
+	err error
+}
+
+//go:generate stringer -type=State
+type State uint
+
+const (
+	SyncUp State = 1<<iota
+	Planning
+	Snapshotting
+	Waiting
+	ErrorWait
+	Stopped
+)
+
+func (s State) sf() state {
+	m := map[State]state{
+		SyncUp:       syncUp,
+		Planning:     plan,
+		Snapshotting: snapshot,
+		Waiting:      wait,
+		ErrorWait:    wait,
+		Stopped: 	   nil,
+	}
+	return m[s]
+}
+
+type updater func(u func(*Snapper)) State
+type state func(a args, u updater) state
+
+type contextKey int
+
+const (
+	contextKeyLog contextKey = 0
+)
+
+type Logger = logger.Logger
+
+func WithLogger(ctx context.Context, log Logger) context.Context {
+	return context.WithValue(ctx, contextKeyLog, log)
+}
+
+func getLogger(ctx context.Context) Logger {
+	if log, ok := ctx.Value(contextKeyLog).(Logger); ok {
+		return log
+	}
+	return logger.NewNullLogger()
+}
+
+func PeriodicFromConfig(g *config.Global, fsf *filters.DatasetMapFilter, in *config.SnapshottingPeriodic) (*Snapper, error) {
+	if in.Prefix == "" {
+		return nil, errors.New("prefix must not be empty")
+	}
+	if in.Interval <= 0 {
+		return nil, errors.New("interval must be positive")
+	}
+
+	args := args{
+		prefix: in.Prefix,
+		interval: in.Interval,
+		fsf: fsf,
+		// ctx and log is set in Run()
+	}
+
+	return &Snapper{state: SyncUp, args: args}, nil
+}
+
+func (s *Snapper) Run(ctx context.Context, snapshotsTaken chan<- struct{}) {
+
+	getLogger(ctx).Debug("start")
+	defer getLogger(ctx).Debug("stop")
+
+	s.args.snapshotsTaken = snapshotsTaken
+	s.args.ctx = ctx
+	s.args.log = getLogger(ctx)
+
+	u := func(u func(*Snapper)) State {
+		s.mtx.Lock()
+		defer s.mtx.Unlock()
+		if u != nil {
+			u(s)
+		}
+		return s.state
+	}
+
+	var st state = syncUp
+
+	for st != nil {
+		pre := u(nil)
+		st = st(s.args, u)
+		post := u(nil)
+		getLogger(ctx).
+			WithField("transition", fmt.Sprintf("%s=>%s", pre, post)).
+			Debug("state transition")
+
+	}
+
+}
+
+func onErr(err error, u updater) state {
+	return u(func(s *Snapper) {
+		s.err = err
+		s.state = ErrorWait
+	}).sf()
+}
+
+func onMainCtxDone(ctx context.Context, u updater) state {
+	return u(func(s *Snapper) {
+		s.err = ctx.Err()
+		s.state = Stopped
+	}).sf()
+}
+
+func syncUp(a args, u updater) state {
+	fss, err := listFSes(a.fsf)
+	if err != nil {
+		return onErr(err, u)
+	}
+	syncPoint, err := findSyncPoint(a.log, fss, a.prefix, a.interval)
+	if err != nil {
+		return onErr(err, u)
+	}
+	u(func(s *Snapper){
+		s.sleepUntil = syncPoint
+	})
+	t := time.NewTimer(syncPoint.Sub(time.Now()))
+	defer t.Stop()
+	select {
+	case <-t.C:
+		return u(func(s *Snapper) {
+			s.state = Planning
+		}).sf()
+	case <-a.ctx.Done():
+		return onMainCtxDone(a.ctx, u)
+	}
+}
+
+func plan(a args, u updater) state {
+	u(func(snapper *Snapper) {
+		snapper.lastInvocation = time.Now()
+	})
+	fss, err := listFSes(a.fsf)
+	if err != nil {
+		return onErr(err, u)
+	}
+
+	plan := make(map[*zfs.DatasetPath]snapProgress, len(fss))
+	for _, fs := range fss {
+		plan[fs] = snapProgress{state: SnapPending}
+	}
+	return u(func(s *Snapper) {
+		s.state = Snapshotting
+		s.plan = plan
+	}).sf()
+}
+
+func snapshot(a args, u updater) state {
+
+	var plan map[*zfs.DatasetPath]snapProgress
+	u(func(snapper *Snapper) {
+		plan = snapper.plan
+	})
+
+	hadErr := false
+	// TODO channel programs -> allow a little jitter?
+	for fs, progress := range plan {
+		suffix := time.Now().In(time.UTC).Format("20060102_150405_000")
+		snapname := fmt.Sprintf("%s%s", a.prefix, suffix)
+
+		l := a.log.
+			WithField("fs", fs.ToString()).
+			WithField("snap", snapname)
+
+		u(func(snapper *Snapper) {
+			progress.name = snapname
+			progress.startAt = time.Now()
+			progress.state = SnapStarted
+		})
+
+		l.Debug("create snapshot")
+		err := zfs.ZFSSnapshot(fs, snapname, false)
+		if err != nil {
+			hadErr = true
+			l.WithError(err).Error("cannot create snapshot")
+		}
+		doneAt := time.Now()
+
+		u(func(snapper *Snapper) {
+			progress.doneAt = doneAt
+			progress.state = SnapDone
+			if err != nil {
+				progress.state = SnapError
+				progress.err = err
+			}
+		})
+	}
+
+	select {
+	case a.snapshotsTaken <- struct{}{}:
+	default:
+		if a.snapshotsTaken != nil {
+			a.log.Warn("callback channel is full, discarding snapshot update event")
+		}
+	}
+
+	return u(func(snapper *Snapper) {
+		if hadErr {
+			snapper.state = ErrorWait
+			snapper.err = errors.New("one or more snapshots could not be created, check logs for details")
+		} else {
+			snapper.state = Waiting
+		}
+	}).sf()
+}
+
+func wait(a args, u updater) state {
+	var sleepUntil time.Time
+	u(func(snapper *Snapper) {
+		lastTick := snapper.lastInvocation
+		snapper.sleepUntil = lastTick.Add(a.interval)
+		sleepUntil = snapper.sleepUntil
+	})
+
+	t := time.NewTimer(sleepUntil.Sub(time.Now()))
+	defer t.Stop()
+
+	select {
+	case <-t.C:
+		return u(func(snapper *Snapper) {
+			snapper.state = Planning
+		}).sf()
+	case <-a.ctx.Done():
+		return onMainCtxDone(a.ctx, u)
+	}
+}
+
+func listFSes(mf *filters.DatasetMapFilter) (fss []*zfs.DatasetPath, err error) {
+	return zfs.ZFSListMapping(mf)
+}
+
+func findSyncPoint(log Logger, fss []*zfs.DatasetPath, prefix string, interval time.Duration) (syncPoint time.Time, err error) {
+	type snapTime struct {
+		ds   *zfs.DatasetPath
+		time time.Time
+	}
+
+	if len(fss) == 0 {
+		return time.Now(), nil
+	}
+
+	snaptimes := make([]snapTime, 0, len(fss))
+
+	now := time.Now()
+
+	log.Debug("examine filesystem state")
+	for _, d := range fss {
+
+		l := log.WithField("fs", d.ToString())
+
+		fsvs, err := zfs.ZFSListFilesystemVersions(d, filters.NewTypedPrefixFilter(prefix, zfs.Snapshot))
+		if err != nil {
+			l.WithError(err).Error("cannot list filesystem versions")
+			continue
+		}
+		if len(fsvs) <= 0 {
+			l.WithField("prefix", prefix).Debug("no filesystem versions with prefix")
+			continue
+		}
+
+		// Sort versions by creation
+		sort.SliceStable(fsvs, func(i, j int) bool {
+			return fsvs[i].CreateTXG < fsvs[j].CreateTXG
+		})
+
+		latest := fsvs[len(fsvs)-1]
+		l.WithField("creation", latest.Creation).
+			Debug("found latest snapshot")
+
+		since := now.Sub(latest.Creation)
+		if since < 0 {
+			l.WithField("snapshot", latest.Name).
+				WithField("creation", latest.Creation).
+				Error("snapshot is from the future")
+			continue
+		}
+		next := now
+		if since < interval {
+			next = latest.Creation.Add(interval)
+		}
+		snaptimes = append(snaptimes, snapTime{d, next})
+	}
+
+	if len(snaptimes) == 0 {
+		snaptimes = append(snaptimes, snapTime{nil, now})
+	}
+
+	sort.Slice(snaptimes, func(i, j int) bool {
+		return snaptimes[i].time.Before(snaptimes[j].time)
+	})
+
+	return snaptimes[0].time, nil
+
+}
+

daemon/snapper/snapper_all.go

@@ -0,0 +1,39 @@
+package snapper
+
+import (
+	"context"
+	"fmt"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/filters"
+)
+
+// FIXME: properly abstract snapshotting:
+//   - split up things that trigger snapshotting from the mechanism
+//     - timer-based trigger (periodic)
+//     - call from control socket (manual)
+//     - mixed modes?
+//   - support a `zrepl snapshot JOBNAME` subcommand for config.SnapshottingManual
+type PeriodicOrManual struct {
+	s *Snapper
+}
+
+func (s *PeriodicOrManual) Run(ctx context.Context, wakeUpCommon chan <- struct{}) {
+	if s.s != nil {
+		s.s.Run(ctx, wakeUpCommon)
+	}
+}
+
+func FromConfig(g *config.Global, fsf *filters.DatasetMapFilter, in config.SnapshottingEnum) (*PeriodicOrManual, error) {
+	switch v := in.Ret.(type) {
+	case *config.SnapshottingPeriodic:
+		snapper, err := PeriodicFromConfig(g, fsf, v)
+		if err != nil {
+			return nil, err
+		}
+		return &PeriodicOrManual{snapper}, nil
+	case *config.SnapshottingManual:
+		return &PeriodicOrManual{}, nil
+	default:
+		return nil, fmt.Errorf("unknown snapshotting type %T", v)
+	}
+}

daemon/snapper/snapstate_string.go

@@ -0,0 +1,29 @@
+// Code generated by "stringer -type=SnapState"; DO NOT EDIT.
+
+package snapper
+
+import "strconv"
+
+const (
+	_SnapState_name_0 = "SnapPendingSnapStarted"
+	_SnapState_name_1 = "SnapDone"
+	_SnapState_name_2 = "SnapError"
+)
+
+var (
+	_SnapState_index_0 = [...]uint8{0, 11, 22}
+)
+
+func (i SnapState) String() string {
+	switch {
+	case 1 <= i && i <= 2:
+		i -= 1
+		return _SnapState_name_0[_SnapState_index_0[i]:_SnapState_index_0[i+1]]
+	case i == 4:
+		return _SnapState_name_1
+	case i == 8:
+		return _SnapState_name_2
+	default:
+		return "SnapState(" + strconv.FormatInt(int64(i), 10) + ")"
+	}
+}

daemon/snapper/state_string.go

@@ -0,0 +1,35 @@
+// Code generated by "stringer -type=State"; DO NOT EDIT.
+
+package snapper
+
+import "strconv"
+
+const (
+	_State_name_0 = "SyncUpPlanning"
+	_State_name_1 = "Snapshotting"
+	_State_name_2 = "Waiting"
+	_State_name_3 = "ErrorWait"
+	_State_name_4 = "Stopped"
+)
+
+var (
+	_State_index_0 = [...]uint8{0, 6, 14}
+)
+
+func (i State) String() string {
+	switch {
+	case 1 <= i && i <= 2:
+		i -= 1
+		return _State_name_0[_State_index_0[i]:_State_index_0[i+1]]
+	case i == 4:
+		return _State_name_1
+	case i == 8:
+		return _State_name_2
+	case i == 16:
+		return _State_name_3
+	case i == 32:
+		return _State_name_4
+	default:
+		return "State(" + strconv.FormatInt(int64(i), 10) + ")"
+	}
+}

daemon/streamrpcconfig/streamrpcconfig.go

@@ -0,0 +1,25 @@
+package streamrpcconfig
+
+import (
+	"github.com/problame/go-streamrpc"
+	"github.com/zrepl/zrepl/config"
+)
+
+func FromDaemonConfig(g *config.Global, in *config.RPCConfig) (*streamrpc.ConnConfig, error) {
+	conf := in
+	if conf == nil {
+		conf = g.RPC
+	}
+	srpcConf := &streamrpc.ConnConfig{
+		RxHeaderMaxLen:       conf.RxHeaderMaxLen,
+		RxStructuredMaxLen:   conf.RxStructuredMaxLen,
+		RxStreamMaxChunkSize: conf.RxStreamChunkMaxLen,
+		TxChunkSize:          conf.TxChunkSize,
+		Timeout: 			  conf.Timeout,
+		SendHeartbeatInterval: conf.SendHeartbeatInterval,
+	}
+	if err := srpcConf.Validate(); err != nil {
+		return nil, err
+	}
+	return srpcConf, nil
+}

daemon/transport/connecter/connect_local.go

@@ -0,0 +1,30 @@
+package connecter
+
+import (
+	"context"
+	"fmt"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/transport/serve"
+	"net"
+)
+
+type LocalConnecter struct {
+	listenerName string
+	clientIdentity string
+}
+
+func LocalConnecterFromConfig(in *config.LocalConnect) (*LocalConnecter, error) {
+	if in.ClientIdentity == "" {
+		return nil, fmt.Errorf("ClientIdentity must not be empty")
+	}
+	if in.ListenerName == "" {
+		return nil, fmt.Errorf("ListenerName must not be empty")
+	}
+	return &LocalConnecter{listenerName: in.ListenerName, clientIdentity: in.ClientIdentity}, nil
+}
+
+func (c *LocalConnecter) Connect(dialCtx context.Context) (conn net.Conn, err error) {
+	l := serve.GetLocalListener(c.listenerName)
+	return l.Connect(dialCtx, c.clientIdentity)
+}
+

daemon/transport/connecter/connect_ssh.go

@@ -0,0 +1,66 @@
+package connecter
+
+import (
+	"context"
+	"github.com/jinzhu/copier"
+	"github.com/pkg/errors"
+	"github.com/problame/go-netssh"
+	"github.com/problame/go-streamrpc"
+	"github.com/zrepl/zrepl/config"
+	"net"
+	"time"
+)
+
+type SSHStdinserverConnecter struct {
+	Host                 string
+	User                 string
+	Port                 uint16
+	IdentityFile         string
+	TransportOpenCommand []string
+	SSHCommand           string
+	Options              []string
+	dialTimeout          time.Duration
+}
+
+var _ streamrpc.Connecter = &SSHStdinserverConnecter{}
+
+func SSHStdinserverConnecterFromConfig(in *config.SSHStdinserverConnect) (c *SSHStdinserverConnecter, err error) {
+
+	c = &SSHStdinserverConnecter{
+		Host:         in.Host,
+		User:         in.User,
+		Port:         in.Port,
+		IdentityFile: in.IdentityFile,
+		SSHCommand:   in.SSHCommand,
+		Options:      in.Options,
+		dialTimeout:  in.DialTimeout,
+	}
+	return
+
+}
+
+type netsshConnToConn struct{ *netssh.SSHConn }
+
+var _ net.Conn = netsshConnToConn{}
+
+func (netsshConnToConn) SetDeadline(dl time.Time) error      { return nil }
+func (netsshConnToConn) SetReadDeadline(dl time.Time) error  { return nil }
+func (netsshConnToConn) SetWriteDeadline(dl time.Time) error { return nil }
+
+func (c *SSHStdinserverConnecter) Connect(dialCtx context.Context) (net.Conn, error) {
+
+	var endpoint netssh.Endpoint
+	if err := copier.Copy(&endpoint, c); err != nil {
+		return nil, errors.WithStack(err)
+	}
+	dialCtx, dialCancel := context.WithTimeout(dialCtx, c.dialTimeout) // context.TODO tied to error handling below
+	defer dialCancel()
+	nconn, err := netssh.Dial(dialCtx, endpoint)
+	if err != nil {
+		if err == context.DeadlineExceeded {
+			err = errors.Errorf("dial_timeout of %s exceeded", c.dialTimeout)
+		}
+		return nil, err
+	}
+	return netsshConnToConn{nconn}, nil
+}

daemon/transport/connecter/connect_tcp.go

@@ -0,0 +1,24 @@
+package connecter
+
+import (
+	"context"
+	"github.com/zrepl/zrepl/config"
+	"net"
+)
+
+type TCPConnecter struct {
+	Address string
+	dialer  net.Dialer
+}
+
+func TCPConnecterFromConfig(in *config.TCPConnect) (*TCPConnecter, error) {
+	dialer := net.Dialer{
+		Timeout: in.DialTimeout,
+	}
+
+	return &TCPConnecter{in.Address, dialer}, nil
+}
+
+func (c *TCPConnecter) Connect(dialCtx context.Context) (conn net.Conn, err error) {
+	return c.dialer.DialContext(dialCtx, "tcp", c.Address)
+}

daemon/transport/connecter/connect_tls.go

@@ -0,0 +1,47 @@
+package connecter
+
+import (
+	"context"
+	"crypto/tls"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/tlsconf"
+	"net"
+)
+
+type TLSConnecter struct {
+	Address   string
+	dialer    net.Dialer
+	tlsConfig *tls.Config
+}
+
+func TLSConnecterFromConfig(in *config.TLSConnect) (*TLSConnecter, error) {
+	dialer := net.Dialer{
+		Timeout: in.DialTimeout,
+	}
+
+	ca, err := tlsconf.ParseCAFile(in.Ca)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse ca file")
+	}
+
+	cert, err := tls.LoadX509KeyPair(in.Cert, in.Key)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse cert/key pair")
+	}
+
+	tlsConfig, err := tlsconf.ClientAuthClient(in.ServerCN, ca, cert)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot build tls config")
+	}
+
+	return &TLSConnecter{in.Address, dialer, tlsConfig}, nil
+}
+
+func (c *TLSConnecter) Connect(dialCtx context.Context) (conn net.Conn, err error) {
+	conn, err = c.dialer.DialContext(dialCtx, "tcp", c.Address)
+	if err != nil {
+		return nil, err
+	}
+	return tls.Client(conn, c.tlsConfig), nil
+}

daemon/transport/connecter/connecter.go

@@ -0,0 +1,84 @@
+package connecter
+
+import (
+	"context"
+	"fmt"
+	"github.com/problame/go-streamrpc"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/streamrpcconfig"
+	"github.com/zrepl/zrepl/daemon/transport"
+	"net"
+	"time"
+)
+
+
+type HandshakeConnecter struct {
+	connecter streamrpc.Connecter
+}
+
+func (c HandshakeConnecter) Connect(ctx context.Context) (net.Conn, error) {
+	conn, err := c.connecter.Connect(ctx)
+	if err != nil {
+		return nil, err
+	}
+	dl, ok := ctx.Deadline()
+	if !ok {
+		dl = time.Now().Add(10 * time.Second) // FIXME constant
+	}
+	if err := transport.DoHandshakeCurrentVersion(conn, dl); err != nil {
+		conn.Close()
+		return nil, err
+	}
+	return conn, nil
+}
+
+
+
+func FromConfig(g *config.Global, in config.ConnectEnum) (*ClientFactory, error) {
+	var (
+		connecter            streamrpc.Connecter
+		errConnecter, errRPC error
+		connConf             *streamrpc.ConnConfig
+	)
+	switch v := in.Ret.(type) {
+	case *config.SSHStdinserverConnect:
+		connecter, errConnecter = SSHStdinserverConnecterFromConfig(v)
+		connConf, errRPC = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.TCPConnect:
+		connecter, errConnecter = TCPConnecterFromConfig(v)
+		connConf, errRPC = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.TLSConnect:
+		connecter, errConnecter = TLSConnecterFromConfig(v)
+		connConf, errRPC = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.LocalConnect:
+		connecter, errConnecter = LocalConnecterFromConfig(v)
+		connConf, errRPC = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	default:
+		panic(fmt.Sprintf("implementation error: unknown connecter type %T", v))
+	}
+
+	if errConnecter != nil {
+		return nil, errConnecter
+	}
+	if errRPC != nil {
+		return nil, errRPC
+	}
+
+	config := streamrpc.ClientConfig{ConnConfig: connConf}
+	if err := config.Validate(); err != nil {
+		return nil, err
+	}
+
+	connecter = HandshakeConnecter{connecter}
+
+	return &ClientFactory{connecter: connecter, config: &config}, nil
+}
+
+type ClientFactory struct {
+	connecter streamrpc.Connecter
+	config    *streamrpc.ClientConfig
+}
+
+func (f ClientFactory) NewClient() (*streamrpc.Client, error) {
+	return streamrpc.NewClient(f.connecter, f.config)
+}

daemon/transport/handshake.go

@@ -0,0 +1,136 @@
+package transport
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"net"
+	"strings"
+	"time"
+	"unicode/utf8"
+)
+
+type HandshakeMessage struct {
+	ProtocolVersion int
+	Extensions []string
+}
+
+func (m *HandshakeMessage) Encode() ([]byte, error) {
+	if m.ProtocolVersion <= 0 || m.ProtocolVersion > 9999 {
+		return nil, fmt.Errorf("protocol version must be in [1, 9999]")
+	}
+	if len(m.Extensions) >= 9999 {
+		return nil, fmt.Errorf("protocol only supports [0, 9999] extensions")
+	}
+	// EXTENSIONS is a count of subsequent \n separated lines that contain protocol extensions
+	var extensions strings.Builder
+	for i, ext := range m.Extensions {
+		if strings.ContainsAny(ext, "\n") {
+			return nil, fmt.Errorf("Extension #%d contains forbidden newline character", i)
+		}
+		if !utf8.ValidString(ext) {
+			return nil, fmt.Errorf("Extension #%d is not valid UTF-8", i)
+		}
+		extensions.WriteString(ext)
+		extensions.WriteString("\n")
+	}
+	withoutLen := fmt.Sprintf("ZREPL_ZFS_REPLICATION PROTOVERSION=%04d EXTENSIONS=%04d\n%s",
+		m.ProtocolVersion, len(m.Extensions), extensions.String())
+	withLen := fmt.Sprintf("%010d %s", len(withoutLen), withoutLen)
+	return []byte(withLen), nil
+}
+
+func (m *HandshakeMessage) DecodeReader(r io.Reader, maxLen int) error {
+	var lenAndSpace [11]byte
+	if _, err := io.ReadFull(r, lenAndSpace[:]); err != nil {
+		return err
+	}
+	if !utf8.Valid(lenAndSpace[:]) {
+		return fmt.Errorf("invalid start of handshake message: not valid UTF-8")
+	}
+	var followLen int
+	n, err := fmt.Sscanf(string(lenAndSpace[:]), "%010d ", &followLen)
+	if n != 1 || err != nil {
+		return fmt.Errorf("could not parse handshake message length")
+	}
+	if followLen > maxLen {
+		return fmt.Errorf("handshake message length exceeds max length (%d vs %d)",
+			followLen, maxLen)
+	}
+
+	var buf bytes.Buffer
+	_, err = io.Copy(&buf, io.LimitReader(r, int64(followLen)))
+	if err != nil {
+		return err
+	}
+
+	var (
+		protoVersion, extensionCount int
+	)
+	n, err = fmt.Fscanf(&buf, "ZREPL_ZFS_REPLICATION PROTOVERSION=%04d EXTENSIONS=%4d\n",
+		&protoVersion, &extensionCount)
+	if n != 2 || err != nil {
+		return fmt.Errorf("could not parse handshake message: %s", err)
+	}
+	if protoVersion < 1 {
+		return fmt.Errorf("invalid protocol version %q", protoVersion)
+	}
+	m.ProtocolVersion = protoVersion
+
+	if extensionCount < 0 {
+		return fmt.Errorf("invalid extension count %q", extensionCount)
+	}
+	if extensionCount == 0 {
+		if buf.Len() != 0 {
+			return fmt.Errorf("unexpected data trailing after header")
+		}
+		m.Extensions = nil
+		return nil
+	}
+	s := buf.String()
+	if strings.Count(s, "\n") != extensionCount {
+		return fmt.Errorf("inconsistent extension count: found %d, header says %d", len(m.Extensions), extensionCount)
+	}
+	exts := strings.Split(s, "\n")
+	if exts[len(exts)-1] != "" {
+		return fmt.Errorf("unexpected data trailing after last extension newline")
+	}
+	m.Extensions = exts[0:len(exts)-1]
+
+	return nil
+}
+
+func DoHandshakeCurrentVersion(conn net.Conn, deadline time.Time) error {
+	// current protocol version is hardcoded here
+	return DoHandshakeVersion(conn, deadline, 1)
+}
+
+func DoHandshakeVersion(conn net.Conn, deadline time.Time, version int) error {
+	ours := HandshakeMessage{
+		ProtocolVersion: version,
+		Extensions: nil,
+	}
+	hsb, err := ours.Encode()
+	if err != nil {
+		return fmt.Errorf("could not encode protocol banner: %s", err)
+	}
+
+	conn.SetDeadline(deadline)
+	_, err = io.Copy(conn, bytes.NewBuffer(hsb))
+	if err != nil {
+		return fmt.Errorf("could not send protocol banner: %s", err)
+	}
+
+	theirs := HandshakeMessage{}
+	if err := theirs.DecodeReader(conn, 16 * 4096); err != nil { // FIXME constant
+		return fmt.Errorf("could not decode protocol banner: %s", err)
+	}
+
+	if theirs.ProtocolVersion != ours.ProtocolVersion {
+		return fmt.Errorf("protocol versions do not match: ours is %d, theirs is %d",
+			ours.ProtocolVersion, theirs.ProtocolVersion)
+	}
+	// ignore extensions, we don't use them
+
+	return nil
+}

daemon/transport/handshake_test.go

@@ -0,0 +1,119 @@
+package transport
+
+import (
+	"bytes"
+	"fmt"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/zrepl/zrepl/util/socketpair"
+	"io"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestHandshakeMessage_Encode(t *testing.T) {
+
+	msg := HandshakeMessage{
+		ProtocolVersion: 2342,
+	}
+
+	encB, err := msg.Encode()
+	require.NoError(t, err)
+	enc := string(encB)
+	t.Logf("enc: %s", enc)
+
+
+
+	assert.False(t, strings.ContainsAny(enc[0:10], " "))
+	assert.True(t, enc[10] == ' ')
+
+	var (
+		headerlen, protoversion, extensionCount int
+	)
+	n, err := fmt.Sscanf(enc, "%010d ZREPL_ZFS_REPLICATION PROTOVERSION=%04d EXTENSIONS=%04d\n",
+		&headerlen, &protoversion, &extensionCount)
+	if n != 3 || (err != nil && err != io.EOF) {
+		t.Fatalf("%v %v", n, err)
+	}
+
+	assert.Equal(t, 2342, protoversion)
+	assert.Equal(t, 0, extensionCount)
+	assert.Equal(t, len(enc)-11, headerlen)
+
+}
+
+func TestHandshakeMessage_Encode_InvalidProtocolVersion(t *testing.T) {
+
+	for _, pv := range []int{-1, 0,  10000, 10001} {
+		t.Logf("testing invalid protocol version = %v", pv)
+		msg := HandshakeMessage{
+			ProtocolVersion: pv,
+		}
+		b, err := msg.Encode()
+		assert.Error(t, err)
+		assert.Nil(t, b)
+	}
+
+}
+
+func TestHandshakeMessage_DecodeReader(t *testing.T) {
+
+	in := HandshakeMessage{
+		2342,
+		[]string{"foo", "bar 2342"},
+	}
+
+	enc, err := in.Encode()
+	require.NoError(t, err)
+
+	out := HandshakeMessage{}
+	err = out.DecodeReader(bytes.NewReader([]byte(enc)), 4 * 4096)
+	assert.NoError(t, err)
+	assert.Equal(t, 2342, out.ProtocolVersion)
+	assert.Equal(t, 2, len(out.Extensions))
+	assert.Equal(t, "foo", out.Extensions[0])
+	assert.Equal(t, "bar 2342", out.Extensions[1])
+
+}
+
+func TestDoHandshakeVersion_ErrorOnDifferentVersions(t *testing.T) {
+	srv, client, err := socketpair.SocketPair()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer srv.Close()
+	defer client.Close()
+
+	srvErrCh := make(chan error)
+	go func() {
+		srvErrCh <- DoHandshakeVersion(srv, time.Now().Add(2*time.Second), 1)
+	}()
+	err = DoHandshakeVersion(client, time.Now().Add(2*time.Second), 2)
+	t.Log(err)
+	assert.Error(t, err)
+	assert.True(t, strings.Contains(err.Error(), "version"))
+
+	srvErr := <-srvErrCh
+	t.Log(srvErr)
+	assert.Error(t, srvErr)
+	assert.True(t, strings.Contains(srvErr.Error(), "version"))
+}
+
+func TestDoHandshakeCurrentVersion(t *testing.T) {
+	srv, client, err := socketpair.SocketPair()
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer srv.Close()
+	defer client.Close()
+
+	srvErrCh := make(chan error)
+	go func() {
+		srvErrCh <- DoHandshakeVersion(srv, time.Now().Add(2*time.Second), 1)
+	}()
+	err = DoHandshakeVersion(client, time.Now().Add(2*time.Second), 1)
+	assert.Nil(t, err)
+	assert.Nil(t, <-srvErrCh)
+
+}

daemon/transport/serve/serve.go

@@ -0,0 +1,147 @@
+package serve
+
+import (
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/transport"
+	"net"
+	"github.com/zrepl/zrepl/daemon/streamrpcconfig"
+	"github.com/problame/go-streamrpc"
+	"context"
+	"github.com/zrepl/zrepl/logger"
+	"github.com/zrepl/zrepl/zfs"
+	"time"
+)
+
+type contextKey int
+
+const contextKeyLog contextKey = 0
+
+type Logger = logger.Logger
+
+func WithLogger(ctx context.Context, log Logger) context.Context {
+	return context.WithValue(ctx, contextKeyLog, log)
+}
+
+func getLogger(ctx context.Context) Logger {
+	if log, ok := ctx.Value(contextKeyLog).(Logger); ok {
+		return log
+	}
+	return logger.NewNullLogger()
+}
+
+type AuthenticatedConn interface {
+	net.Conn
+	// ClientIdentity must be a string that satisfies ValidateClientIdentity
+	ClientIdentity() string
+}
+
+// A client identity must be a single component in a ZFS filesystem path
+func ValidateClientIdentity(in string) (err error) {
+	path, err := zfs.NewDatasetPath(in)
+	if err != nil {
+		return err
+	}
+	if path.Length() != 1 {
+		return errors.New("client identity must be a single path comonent (not empty, no '/')")
+	}
+	return nil
+}
+
+type authConn struct {
+	net.Conn
+	clientIdentity string
+}
+
+var _ AuthenticatedConn = authConn{}
+
+func (c authConn) ClientIdentity() string {
+	if err := ValidateClientIdentity(c.clientIdentity); err != nil {
+		panic(err)
+	}
+	return c.clientIdentity
+}
+
+// like net.Listener, but with an AuthenticatedConn instead of net.Conn
+type AuthenticatedListener interface {
+	Addr() (net.Addr)
+	Accept(ctx context.Context) (AuthenticatedConn, error)
+	Close() error
+}
+
+type ListenerFactory interface {
+	Listen() (AuthenticatedListener, error)
+}
+
+type HandshakeListenerFactory struct {
+	lf ListenerFactory
+}
+
+func (lf HandshakeListenerFactory) Listen() (AuthenticatedListener, error) {
+	l, err := lf.lf.Listen()
+	if err != nil {
+		return nil, err
+	}
+	return HandshakeListener{l}, nil
+}
+
+type HandshakeListener struct {
+	l AuthenticatedListener
+}
+
+func (l HandshakeListener) Addr() (net.Addr) { return l.l.Addr() }
+
+func (l HandshakeListener) Close() error { return l.l.Close() }
+
+func (l HandshakeListener) Accept(ctx context.Context) (AuthenticatedConn, error) {
+	conn, err := l.l.Accept(ctx)
+	if err != nil {
+		return nil, err
+	}
+	dl, ok := ctx.Deadline()
+	if !ok {
+		dl = time.Now().Add(10*time.Second) // FIXME constant
+	}
+	if err := transport.DoHandshakeCurrentVersion(conn, dl); err != nil {
+		conn.Close()
+		return nil, err
+	}
+	return conn, nil
+}
+
+func FromConfig(g *config.Global, in config.ServeEnum) (lf ListenerFactory, conf *streamrpc.ConnConfig, _ error) {
+
+	var (
+		lfError, rpcErr error
+	)
+	switch v := in.Ret.(type) {
+	case *config.TCPServe:
+		lf, lfError = TCPListenerFactoryFromConfig(g, v)
+		conf, rpcErr = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.TLSServe:
+		lf, lfError = TLSListenerFactoryFromConfig(g, v)
+		conf, rpcErr = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.StdinserverServer:
+		lf, lfError = MultiStdinserverListenerFactoryFromConfig(g, v)
+		conf, rpcErr = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	case *config.LocalServe:
+		lf, lfError = LocalListenerFactoryFromConfig(g, v)
+		conf, rpcErr = streamrpcconfig.FromDaemonConfig(g, v.RPC)
+	default:
+		return nil, nil, errors.Errorf("internal error: unknown serve type %T", v)
+	}
+
+	if lfError != nil {
+		return nil, nil, lfError
+	}
+	if rpcErr != nil {
+		return nil, nil, rpcErr
+	}
+
+	lf = HandshakeListenerFactory{lf}
+
+	return lf, conf, nil
+
+}
+
+

daemon/transport/serve/serve_local.go

@@ -0,0 +1,187 @@
+package serve
+
+import (
+	"context"
+	"fmt"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/util/socketpair"
+	"net"
+	"sync"
+)
+
+var localListeners struct {
+	m map[string]*LocalListener // listenerName -> listener
+	init sync.Once
+	mtx sync.Mutex
+}
+
+func GetLocalListener(listenerName string) (*LocalListener) {
+
+	localListeners.init.Do(func() {
+		localListeners.m = make(map[string]*LocalListener)
+	})
+
+	localListeners.mtx.Lock()
+	defer localListeners.mtx.Unlock()
+
+	l, ok := localListeners.m[listenerName]
+	if !ok {
+		l = newLocalListener()
+		localListeners.m[listenerName] = l
+	}
+	return l
+
+}
+
+type connectRequest struct {
+	clientIdentity string
+	callback chan connectResult
+}
+
+type connectResult struct {
+	conn net.Conn
+	err error
+}
+
+type LocalListener struct {
+	connects chan connectRequest
+}
+
+func newLocalListener() *LocalListener {
+	return &LocalListener{
+		connects: make(chan connectRequest),
+	}
+}
+
+// Connect to the LocalListener from a client with identity clientIdentity
+func (l *LocalListener) Connect(dialCtx context.Context, clientIdentity string) (conn net.Conn, err error) {
+
+	// place request
+	req := connectRequest{
+		clientIdentity: clientIdentity,
+		callback: make(chan connectResult),
+	}
+	select {
+	case l.connects <- req:
+	case <-dialCtx.Done():
+		return nil, dialCtx.Err()
+	}
+
+	// wait for listener response
+	select {
+	case connRes := <- req.callback:
+		conn, err = connRes.conn, connRes.err
+	case <-dialCtx.Done():
+		close(req.callback) // sending to the channel afterwards will panic, the listener has to catch this
+		conn, err = nil, dialCtx.Err()
+	}
+
+	return conn, err
+}
+
+type localAddr struct {
+	S string
+}
+
+func (localAddr) Network() string { return "local" }
+
+func (a localAddr) String() string { return a.S }
+
+func (l *LocalListener) Addr() (net.Addr) { return localAddr{"<listening>"} }
+
+type localConn struct {
+	net.Conn
+	clientIdentity string
+}
+
+func (l localConn) ClientIdentity() string { return l.clientIdentity }
+
+func (l *LocalListener) Accept(ctx context.Context) (AuthenticatedConn, error) {
+	respondToRequest := func(req connectRequest, res connectResult) (err error) {
+		getLogger(ctx).
+			WithField("res.conn", res.conn).WithField("res.err", res.err).
+			Debug("responding to client request")
+		defer func() {
+			errv := recover()
+			getLogger(ctx).WithField("recover_err", errv).
+				Debug("panic on send to client callback, likely a legitimate client-side timeout")
+		}()
+		select {
+		case req.callback <- res:
+			err = nil
+		default:
+			err = fmt.Errorf("client-provided callback did block on send")
+		}
+		close(req.callback)
+		return err
+	}
+
+	getLogger(ctx).Debug("waiting for local client connect requests")
+	var req connectRequest
+	select {
+	case req = <-l.connects:
+	case <-ctx.Done():
+		return nil, ctx.Err()
+	}
+
+	getLogger(ctx).WithField("client_identity", req.clientIdentity).Debug("got connect request")
+	if req.clientIdentity == "" {
+		res := connectResult{nil, fmt.Errorf("client identity must not be empty")}
+		if err := respondToRequest(req, res); err != nil {
+			return nil, err
+		}
+		return nil, fmt.Errorf("client connected with empty client identity")
+	}
+
+	getLogger(ctx).Debug("creating socketpair")
+	left, right, err := socketpair.SocketPair()
+	if err != nil {
+		res := connectResult{nil, fmt.Errorf("server error: %s", err)}
+		if respErr := respondToRequest(req, res); respErr != nil {
+			// returning the socketpair error properly is more important than the error sent to the client
+			getLogger(ctx).WithError(respErr).Error("error responding to client")
+		}
+		return nil, err
+	}
+
+	getLogger(ctx).Debug("responding with left side of socketpair")
+	res := connectResult{left, nil}
+	if err := respondToRequest(req, res); err != nil {
+		getLogger(ctx).WithError(err).Error("error responding to client")
+		if err := left.Close(); err != nil {
+			getLogger(ctx).WithError(err).Error("cannot close left side of socketpair")
+		}
+		if err := right.Close(); err != nil {
+			getLogger(ctx).WithError(err).Error("cannot close right side of socketpair")
+		}
+		return nil, err
+	}
+
+	return localConn{right, req.clientIdentity}, nil
+}
+
+func (l *LocalListener) Close() error {
+	// FIXME: make sure concurrent Accepts return with error, and further Accepts return that error, too
+	// Example impl: for each accept, do context.WithCancel, and store the cancel in a list
+	// When closing, set a member variable to state=closed, make sure accept will exit early
+	// and then call all cancels in the list
+	// The code path from Accept entry over check if state=closed to list entry must be protected by a mutex.
+	return nil
+}
+
+type LocalListenerFactory struct {
+	listenerName string
+}
+
+func LocalListenerFactoryFromConfig(g *config.Global, in *config.LocalServe) (f *LocalListenerFactory, err error) {
+	if in.ListenerName == "" {
+		return nil, fmt.Errorf("ListenerName must not be empty")
+	}
+	return &LocalListenerFactory{listenerName: in.ListenerName}, nil
+}
+
+
+func (lf *LocalListenerFactory) Listen() (AuthenticatedListener, error) {
+	return GetLocalListener(lf.listenerName), nil
+}
+

daemon/transport/serve/serve_stdinserver.go

@@ -0,0 +1,158 @@
+package serve
+
+import (
+	"github.com/problame/go-netssh"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/daemon/nethelpers"
+	"io"
+	"net"
+	"path"
+	"time"
+	"context"
+	"github.com/pkg/errors"
+	"sync/atomic"
+)
+
+type StdinserverListenerFactory struct {
+	ClientIdentities []string
+	Sockdir string
+}
+
+func MultiStdinserverListenerFactoryFromConfig(g *config.Global, in *config.StdinserverServer) (f *multiStdinserverListenerFactory, err error) {
+
+	for _, ci := range in.ClientIdentities {
+		if err := ValidateClientIdentity(ci); err != nil {
+			return nil, errors.Wrapf(err, "invalid client identity %q", ci)
+		}
+	}
+
+	f = &multiStdinserverListenerFactory{
+		ClientIdentities: in.ClientIdentities,
+		Sockdir: g.Serve.StdinServer.SockDir,
+	}
+
+	return
+}
+
+type multiStdinserverListenerFactory struct {
+	ClientIdentities []string
+	Sockdir string
+}
+
+func (f *multiStdinserverListenerFactory) Listen() (AuthenticatedListener, error) {
+	return multiStdinserverListenerFromClientIdentities(f.Sockdir, f.ClientIdentities)
+}
+
+type multiStdinserverAcceptRes struct {
+	conn AuthenticatedConn
+	err error
+}
+
+type MultiStdinserverListener struct {
+	listeners []*stdinserverListener
+	accepts chan multiStdinserverAcceptRes
+	closed int32
+}
+
+// client identities must be validated
+func multiStdinserverListenerFromClientIdentities(sockdir string, cis []string) (*MultiStdinserverListener, error) {
+	listeners := make([]*stdinserverListener, 0, len(cis))
+	var err error
+	for _, ci := range cis {
+		sockpath := path.Join(sockdir, ci)
+		l  := &stdinserverListener{clientIdentity: ci}
+		if err = nethelpers.PreparePrivateSockpath(sockpath); err != nil {
+			break
+		}
+		if l.l, err = netssh.Listen(sockpath); err != nil {
+			break
+		}
+		listeners = append(listeners, l)
+	}
+	if err != nil {
+		for _, l := range listeners {
+			l.Close() // FIXME error reporting?
+		}
+		return nil, err
+	}
+	return &MultiStdinserverListener{listeners: listeners}, nil
+}
+
+func (m *MultiStdinserverListener) Accept(ctx context.Context) (AuthenticatedConn, error){
+
+	if m.accepts == nil {
+		m.accepts = make(chan multiStdinserverAcceptRes, len(m.listeners))
+		for i := range m.listeners {
+			go func(i int) {
+				for atomic.LoadInt32(&m.closed) == 0 {
+					conn, err := m.listeners[i].Accept(context.TODO())
+					m.accepts <- multiStdinserverAcceptRes{conn, err}
+				}
+			}(i)
+		}
+	}
+
+	res := <- m.accepts
+	return res.conn, res.err
+
+}
+
+func (m *MultiStdinserverListener) Addr() (net.Addr) {
+	return netsshAddr{}
+}
+
+func (m *MultiStdinserverListener) Close() error {
+	atomic.StoreInt32(&m.closed, 1)
+	var oneErr error
+	for _, l := range m.listeners {
+		if err := l.Close(); err != nil && oneErr == nil {
+			oneErr = err
+		}
+	}
+	return oneErr
+}
+
+// a single stdinserverListener (part of multiStinserverListener)
+type stdinserverListener struct {
+	l *netssh.Listener
+	clientIdentity string
+}
+
+func (l stdinserverListener) Addr() net.Addr {
+	return netsshAddr{}
+}
+
+func (l stdinserverListener) Accept(ctx context.Context) (AuthenticatedConn, error) {
+	c, err := l.l.Accept()
+	if err != nil {
+		return nil, err
+	}
+	return netsshConnToNetConnAdatper{c, l.clientIdentity}, nil
+}
+
+func (l stdinserverListener) Close() (err error) {
+	return l.l.Close()
+}
+
+type netsshAddr struct{}
+
+func (netsshAddr) Network() string { return "netssh" }
+func (netsshAddr) String() string  { return "???" }
+
+type netsshConnToNetConnAdatper struct {
+	io.ReadWriteCloser // works for both netssh.SSHConn and netssh.ServeConn
+	clientIdentity string
+}
+
+func (a netsshConnToNetConnAdatper) ClientIdentity() string { return a.clientIdentity }
+
+func (netsshConnToNetConnAdatper) LocalAddr() net.Addr { return netsshAddr{} }
+
+func (netsshConnToNetConnAdatper) RemoteAddr() net.Addr { return netsshAddr{} }
+
+// FIXME log warning once!
+func (netsshConnToNetConnAdatper) SetDeadline(t time.Time) error { return nil }
+
+func (netsshConnToNetConnAdatper) SetReadDeadline(t time.Time) error { return nil }
+
+func (netsshConnToNetConnAdatper) SetWriteDeadline(t time.Time) error { return nil }

daemon/transport/serve/serve_tcp.go

@@ -0,0 +1,91 @@
+package serve
+
+import (
+	"github.com/zrepl/zrepl/config"
+	"net"
+	"github.com/pkg/errors"
+	"context"
+)
+
+type TCPListenerFactory struct {
+	address *net.TCPAddr
+	clientMap *ipMap
+}
+
+type ipMapEntry struct {
+	ip net.IP
+	ident string
+}
+
+type ipMap struct {
+	entries []ipMapEntry
+}
+
+func ipMapFromConfig(clients map[string]string) (*ipMap, error) {
+	entries := make([]ipMapEntry, 0, len(clients))
+	for clientIPString, clientIdent := range clients {
+		clientIP := net.ParseIP(clientIPString)
+		if clientIP == nil {
+			return nil, errors.Errorf("cannot parse client IP %q", clientIPString)
+		}
+		if err := ValidateClientIdentity(clientIdent); err != nil {
+			return nil, errors.Wrapf(err,"invalid client identity for IP %q", clientIPString)
+		}
+		entries = append(entries, ipMapEntry{clientIP, clientIdent})
+	}
+	return &ipMap{entries: entries}, nil
+}
+
+func (m *ipMap) Get(ip net.IP) (string, error) {
+	for _, e := range m.entries {
+		if e.ip.Equal(ip) {
+			return e.ident, nil
+		}
+	}
+	return "", errors.Errorf("no identity mapping for client IP %s", ip)
+}
+
+func TCPListenerFactoryFromConfig(c *config.Global, in *config.TCPServe) (*TCPListenerFactory, error) {
+	addr, err := net.ResolveTCPAddr("tcp", in.Listen)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse listen address")
+	}
+	clientMap, err := ipMapFromConfig(in.Clients)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse client IP map")
+	}
+	lf := &TCPListenerFactory{
+		address: addr,
+		clientMap: clientMap,
+	}
+	return lf, nil
+}
+
+func (f *TCPListenerFactory) Listen() (AuthenticatedListener, error) {
+	l, err := net.ListenTCP("tcp", f.address)
+	if err != nil {
+		return nil, err
+	}
+	return &TCPAuthListener{l, f.clientMap}, nil
+}
+
+type TCPAuthListener struct {
+	*net.TCPListener
+	clientMap *ipMap
+}
+
+func (f *TCPAuthListener) Accept(ctx context.Context) (AuthenticatedConn, error) {
+	nc, err := f.TCPListener.Accept()
+	if err != nil {
+		return nil, err
+	}
+	clientIP := nc.RemoteAddr().(*net.TCPAddr).IP
+	clientIdent, err := f.clientMap.Get(clientIP)
+	if err != nil {
+		getLogger(ctx).WithField("ip", clientIP).Error("client IP not in client map")
+		nc.Close()
+		return nil, err
+	}
+	return authConn{nc, clientIdent}, nil
+}
+

daemon/transport/serve/serve_tls.go

@@ -0,0 +1,83 @@
+package serve
+
+import (
+	"crypto/tls"
+	"crypto/x509"
+	"fmt"
+	"github.com/pkg/errors"
+	"github.com/zrepl/zrepl/config"
+	"github.com/zrepl/zrepl/tlsconf"
+	"net"
+	"time"
+	"context"
+)
+
+type TLSListenerFactory struct {
+	address          string
+	clientCA         *x509.CertPool
+	serverCert       tls.Certificate
+	handshakeTimeout time.Duration
+	clientCNs map[string]struct{}
+}
+
+func TLSListenerFactoryFromConfig(c *config.Global, in *config.TLSServe) (lf *TLSListenerFactory, err error) {
+	lf = &TLSListenerFactory{
+		address: in.Listen,
+		handshakeTimeout: in.HandshakeTimeout,
+	}
+
+	if in.Ca == "" || in.Cert == "" || in.Key == "" {
+		return nil, errors.New("fields 'ca', 'cert' and 'key'must be specified")
+	}
+
+	lf.clientCA, err = tlsconf.ParseCAFile(in.Ca)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse ca file")
+	}
+
+	lf.serverCert, err = tls.LoadX509KeyPair(in.Cert, in.Key)
+	if err != nil {
+		return nil, errors.Wrap(err, "cannot parse cer/key pair")
+	}
+
+	lf.clientCNs = make(map[string]struct{}, len(in.ClientCNs))
+	for i, cn := range in.ClientCNs {
+		if err := ValidateClientIdentity(cn); err != nil {
+			return nil, errors.Wrapf(err, "unsuitable client_cn #%d %q", i, cn)
+		}
+		// dupes are ok fr now
+		lf.clientCNs[cn] = struct{}{}
+	}
+
+	return lf, nil
+}
+
+func (f *TLSListenerFactory) Listen() (AuthenticatedListener, error) {
+	l, err := net.Listen("tcp", f.address)
+	if err != nil {
+		return nil, err
+	}
+	tl := tlsconf.NewClientAuthListener(l, f.clientCA, f.serverCert, f.handshakeTimeout)
+	return tlsAuthListener{tl, f.clientCNs}, nil
+}
+
+type tlsAuthListener struct {
+	*tlsconf.ClientAuthListener
+	clientCNs map[string]struct{}
+}
+
+func (l tlsAuthListener) Accept(ctx context.Context) (AuthenticatedConn, error) {
+	c, cn, err := l.ClientAuthListener.Accept()
+	if err != nil {
+		return nil, err
+	}
+	if _, ok := l.clientCNs[cn]; !ok {
+		if err := c.Close(); err != nil {
+			getLogger(ctx).WithError(err).Error("error closing connection with unauthorized common name")
+		}
+		return nil, fmt.Errorf("unauthorized client common name %q from %s", cn, c.RemoteAddr())
+	}
+	return authConn{c, cn}, nil
+}
+
+

docs/.gitignore

@@ -1 +1,2 @@
-public_git
+_build/
+public_git/

docs/Makefile

@@ -0,0 +1,20 @@
+# Minimal makefile for Sphinx documentation
+#
+
+# You can set these variables from the command line.
+SPHINXOPTS    =
+SPHINXBUILD   = sphinx-build
+SPHINXPROJ    = zrepl
+SOURCEDIR     = .
+BUILDDIR      = _build
+
+# Put it first so that "make" without argument is like "make help".
+help:
+	@$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" -c sphinxconf $(SPHINXOPTS) $(O)
+
+.PHONY: help Makefile
+
+# Catch-all target: route all unknown targets to Sphinx using the new
+# "make mode" option.  $(O) is meant as a shortcut for $(SPHINXOPTS).
+%: Makefile
+	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" -c sphinxconf $(SPHINXOPTS) $(O)

docs/_static/zrepl.svg

@@ -0,0 +1,147 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg
+   xmlns:dc="http://purl.org/dc/elements/1.1/"
+   xmlns:cc="http://creativecommons.org/ns#"
+   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
+   xmlns:svg="http://www.w3.org/2000/svg"
+   xmlns="http://www.w3.org/2000/svg"
+   xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
+   xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
+   id="svg8"
+   version="1.1"
+   viewBox="0 0 70.555558 24.694445"
+   height="70pt"
+   width="200pt"
+   sodipodi:docname="zrepl.svg"
+   inkscape:version="0.92.2 5c3e80d, 2017-08-06">
+  <sodipodi:namedview
+     pagecolor="#ffffff"
+     bordercolor="#666666"
+     borderopacity="1"
+     objecttolerance="10"
+     gridtolerance="10"
+     guidetolerance="10"
+     inkscape:pageopacity="0"
+     inkscape:pageshadow="2"
+     inkscape:window-width="2560"
+     inkscape:window-height="1405"
+     id="namedview10"
+     showgrid="false"
+     inkscape:zoom="3.1350001"
+     inkscape:cx="6.2268563"
+     inkscape:cy="-2.6953765"
+     inkscape:window-x="0"
+     inkscape:window-y="35"
+     inkscape:window-maximized="1"
+     inkscape:current-layer="layer1" />
+  <defs
+     id="defs2" />
+  <metadata
+     id="metadata5">
+    <rdf:RDF>
+      <cc:Work
+         rdf:about="">
+        <dc:format>image/svg+xml</dc:format>
+        <dc:type
+           rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
+        <dc:title />
+      </cc:Work>
+    </rdf:RDF>
+  </metadata>
+  <g
+     transform="translate(0,-272.30556)"
+     id="layer1">
+    <g
+       aria-label="zrepl"
+       transform="matrix(0.26458334,0,0,0.26458334,0.52916655,273.36389)"
+       style="font-style:normal;font-variant:normal;font-weight:bold;font-stretch:normal;font-size:40px;line-height:25px;font-family:monospace;-inkscape-font-specification:'monospace, Bold';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:center;letter-spacing:0px;word-spacing:0px;writing-mode:lr-tb;text-anchor:middle;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:1px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+       id="flowRoot829">
+      <path
+         d="m 74.504883,26.943359 11.523437,-13.125 H 74.973633 V 9.5410156 h 17.480469 v 4.4726564 l -11.503907,13.125 h 11.503907 v 4.277344 H 74.504883 Z"
+         style="stroke:#f8f8f8;stroke-opacity:1;fill:#f8f8f8;fill-opacity:1"
+         id="path873" />
+      <path
+         d="m 100.11035,9.5410156 h 5.15625 l 0.56641,3.3984374 q 0.83984,-1.875 2.57812,-2.910156 1.73828,-1.0156251 4.00391,-1.0156251 1.26953,0 2.36328,0.2734375 1.25,0.3515625 2.12891,0.8398436 v 5.449219 q -0.9375,-0.859375 -2.1875,-1.289063 -1.23047,-0.429687 -2.75391,-0.429687 -1.81641,0 -3.18359,0.625 -1.36719,0.605469 -2.10938,1.835937 -0.19531,0.3125 -0.37109,0.761719 -0.17578,0.449219 -0.27344,1.054688 -0.0977,0.566406 -0.15625,1.308593 -0.0391,0.742188 -0.0391,1.894532 v 10.078125 h -5.72266 z"
+         style="stroke:#f8f8f8;stroke-opacity:1;fill:#f8f8f8;fill-opacity:1"
+         id="path875" />
+      <path
+         d="m 132.4541,31.982422 q -2.85156,0 -4.9414,-0.722656 -2.07032,-0.742188 -3.53516,-2.226563 -2.91016,-2.96875 -2.91016,-8.4375 0,-5.371094 2.79297,-8.476562 2.79297,-3.1054691 7.65625,-3.1054691 3.10547,-0.097656 5.42969,1.2500001 2.34375,1.347656 3.59375,3.867187 1.26953,2.519532 1.17187,5.878907 v 2.324218 H 126.8877 q 0.0195,2.578125 1.52343,3.847657 1.50391,1.269531 4.47266,1.269531 1.93359,0 3.90625,-0.566406 0.89844,-0.273438 1.89453,-0.703125 0.99609,-0.429688 2.10938,-1.09375 v 5.253906 q -1.99219,0.820312 -4.0625,1.230469 -1.05469,0.214843 -2.14844,0.3125 -1.07422,0.09766 -2.12891,0.09766 z m 3.4961,-13.945313 q -0.0195,-2.285156 -1.15235,-3.4375 -1.11328,-1.152343 -3.28125,-1.152343 -2.01172,0 -3.18359,1.191406 -1.17188,1.210937 -1.38672,3.417969 z"
+         style="stroke:#f8f8f8;stroke-opacity:1;fill:#f8f8f8;fill-opacity:1"
+         id="path877" />
+      <path
+         d="m 146.45801,9.5410156 h 5.13672 l 0.5664,3.2812504 q 0.89844,-1.894532 2.36328,-2.8515629 1.42579,-0.9570312 3.45704,-0.9570312 3.71093,0 5.80078,2.9882811 1.05468,1.523438 1.5625,3.691406 0.50781,2.148438 0.50781,4.824219 0,2.851563 -0.54688,4.980469 -0.54687,2.109375 -1.54296,3.515625 -1.07422,1.523437 -2.59766,2.246094 -1.50391,0.722656 -3.28125,0.722656 -0.91797,0 -1.73828,-0.214844 -0.80078,-0.195312 -1.54297,-0.703125 -0.625,-0.410156 -1.28906,-1.09375 -0.64453,-0.703125 -1.15235,-1.640625 v 11.40625 h -5.70312 z m 9.70703,17.6757814 q 1.85547,0 2.91016,-1.796875 1.05468,-1.777344 1.05468,-4.960938 0,-3.164062 -1.05468,-4.960937 -1.05469,-1.796875 -2.91016,-1.796875 -1.83594,0 -2.92969,1.796875 -1.07422,1.777344 -1.07422,4.960937 0,3.183594 1.07422,4.960938 1.09375,1.796875 2.92969,1.796875 z"
+         style="stroke:#f8f8f8;stroke-opacity:1;fill:#f8f8f8;fill-opacity:1"
+         id="path879" />
+      <path
+         d="m 182.98145,31.572266 q -4.0625,0 -5.85938,-2.109375 -1.77734,-2.070313 -1.77734,-7.011719 V 5.4199219 h -5.80078 V 1.0253906 h 11.52343 V 22.451172 q 0,2.558594 0.80078,3.652344 0.78125,1.074218 2.71485,1.074218 h 4.57031 v 4.394532 z"
+         style="stroke:#f8f8f8;stroke-opacity:1;fill:#f8f8f8;fill-opacity:1"
+         id="path881" />
+    </g>
+    <path
+       inkscape:connector-curvature="0"
+       id="path837"
+       d="M 26.170418,286.30982 H 43.35751"
+       style="font-style:normal;font-weight:normal;font-size:medium;line-height:25px;font-family:sans-serif;letter-spacing:0px;word-spacing:0px;fill:#f8f8f8;fill-opacity:1;fill-rule:evenodd;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1" />
+    <g
+       aria-label="ZFS replication"
+       style="font-style:normal;font-weight:normal;font-size:10.58333397px;line-height:6.61458349px;font-family:sans-serif;letter-spacing:0px;word-spacing:0px;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-linecap:butt;stroke-linejoin:miter;stroke-opacity:1"
+       id="text820">
+      <path
+         d="m 14.758548,294.66467 h -3.493327 v -0.47887 l 2.652724,-3.99286 h -2.570041 v -0.565 h 3.338297 v 0.47887 l -2.652723,3.99287 h 2.72507 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path844" />
+      <path
+         d="M 16.353627,294.66467 H 15.71973 v -5.03673 h 2.814643 v 0.55811 h -2.180746 v 1.78456 h 2.046387 v 0.55466 h -2.046387 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path846" />
+      <path
+         d="m 22.241296,293.31764 q 0,0.65801 -0.475423,1.03697 -0.475424,0.37896 -1.322917,0.37896 -0.895725,0 -1.384929,-0.23771 v -0.60633 q 0.320394,0.1378 0.699355,0.21704 0.382406,0.0792 0.713135,0.0792 0.564995,0 0.850938,-0.21704 0.285943,-0.21704 0.285943,-0.596 0,-0.25149 -0.103353,-0.41341 -0.103352,-0.16192 -0.347954,-0.30317 -0.241157,-0.14125 -0.733806,-0.32039 -0.699354,-0.25494 -0.999077,-0.59945 -0.299724,-0.34795 -0.299724,-0.89572 0,-0.58911 0.440973,-0.93707 0.444417,-0.34795 1.167887,-0.34795 0.754476,0 1.388373,0.28249 l -0.19637,0.54433 q -0.647678,-0.26872 -1.205783,-0.26872 -0.447863,0 -0.7028,0.19293 -0.251492,0.19292 -0.251492,0.54088 0,0.24804 0.09646,0.41341 0.09991,0.16192 0.316949,0.29628 0.220486,0.13436 0.689019,0.30661 0.56155,0.20671 0.837158,0.40308 0.275608,0.19292 0.406521,0.44786 0.130914,0.25149 0.130914,0.60289 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path848" />
+      <path
+         d="m 26.768151,290.81306 q 0.251492,0 0.444418,0.0413 l -0.07579,0.57188 q -0.21015,-0.0482 -0.409966,-0.0482 -0.299723,0 -0.558105,0.16537 -0.254937,0.16536 -0.399631,0.45819 -0.144694,0.28939 -0.144694,0.64424 v 2.01882 h -0.620118 v -3.78271 h 0.509874 l 0.0689,0.69246 h 0.02756 q 0.217041,-0.37551 0.513319,-0.56844 0.296279,-0.19292 0.644233,-0.19292 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path850" />
+      <path
+         d="m 29.465661,294.73357 q -0.850939,0 -1.336697,-0.51331 -0.482313,-0.51677 -0.482313,-1.41938 0,-0.90951 0.451307,-1.44694 0.451308,-0.54088 1.219564,-0.54088 0.713135,0 1.136881,0.46164 0.423747,0.4582 0.423747,1.24368 v 0.37551 h -2.590712 q 0.01723,0.64424 0.327284,0.97841 0.310059,0.33418 0.8785,0.33418 0.303168,0 0.575331,-0.0517 0.272162,-0.0551 0.637342,-0.21015 v 0.54432 q -0.313503,0.13436 -0.589111,0.18948 -0.275608,0.0551 -0.651123,0.0551 z m -0.155029,-3.41408 q -0.444418,0 -0.7028,0.28594 -0.258382,0.28594 -0.306613,0.79582 h 1.925808 q -0.0069,-0.53055 -0.244602,-0.80616 -0.237711,-0.2756 -0.671793,-0.2756 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path852" />
+      <path
+         d="m 33.630782,294.73357 q -0.761366,0 -1.171333,-0.54777 h -0.04134 l 0.01378,0.14125 q 0.02756,0.26872 0.02756,0.47887 v 1.55374 h -0.620117 v -5.4777 h 0.509874 l 0.08268,0.51676 h 0.02756 q 0.220486,-0.31006 0.509874,-0.44786 0.289388,-0.1378 0.668349,-0.1378 0.73725,0 1.143772,0.51332 0.409966,0.50987 0.409966,1.44005 0,0.92673 -0.409966,1.44694 -0.409967,0.5202 -1.150662,0.5202 z m -0.103353,-3.4003 q -0.561551,0 -0.813043,0.31694 -0.248047,0.31695 -0.254937,0.99564 v 0.12058 q 0,0.76825 0.254937,1.11276 0.254937,0.34107 0.826823,0.34107 0.475423,0 0.740696,-0.38586 0.268717,-0.38585 0.268717,-1.07487 0,-0.69246 -0.268717,-1.05764 -0.265273,-0.36862 -0.754476,-0.36862 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path854" />
+      <path
+         d="m 36.800269,294.66467 h -0.620117 v -5.36057 h 0.620117 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path856" />
+      <path
+         d="m 38.619281,294.66467 h -0.620117 v -3.78271 h 0.620117 z m -0.668348,-4.80591 q 0,-0.2067 0.103352,-0.29972 0.106798,-0.093 0.261828,-0.093 0.144694,0 0.251492,0.093 0.110243,0.093 0.110243,0.29972 0,0.20327 -0.110243,0.29973 -0.106798,0.0965 -0.251492,0.0965 -0.15503,0 -0.261828,-0.0965 -0.103352,-0.0965 -0.103352,-0.29973 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path858" />
+      <path
+         d="m 41.334016,294.73357 q -0.816488,0 -1.27124,-0.49953 -0.454753,-0.50299 -0.454753,-1.44005 0,-0.95085 0.458198,-1.46417 0.461642,-0.51676 1.312581,-0.51676 0.275608,0 0.544325,0.0586 0.272163,0.0551 0.447863,0.1447 l -0.189481,0.51676 q -0.478868,-0.17914 -0.816487,-0.17914 -0.571886,0 -0.844049,0.36173 -0.272162,0.35829 -0.272162,1.07143 0,0.68557 0.272162,1.05075 0.272163,0.36173 0.806153,0.36173 0.499538,0 0.981852,-0.22048 v 0.55121 q -0.392741,0.20326 -0.974962,0.20326 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path860" />
+      <path
+         d="m 45.543923,294.66467 -0.120578,-0.53743 h -0.02756 q -0.282498,0.35484 -0.564996,0.48231 -0.279052,0.12402 -0.706244,0.12402 -0.558106,0 -0.875055,-0.29283 -0.316948,-0.29283 -0.316948,-0.82682 0,-1.15066 1.815565,-1.20578 l 0.640788,-0.0241 v -0.22393 q 0,-0.43064 -0.186035,-0.6339 -0.186035,-0.2067 -0.596002,-0.2067 -0.299723,0 -0.568441,0.0896 -0.265272,0.0896 -0.499538,0.19981 l -0.189481,-0.46508 q 0.285943,-0.15159 0.623563,-0.23772 0.337619,-0.0861 0.668348,-0.0861 0.685574,0 1.019748,0.30316 0.334175,0.30317 0.334175,0.96463 v 2.57693 z m -1.27813,-0.43063 q 0.520209,0 0.816487,-0.27906 0.299724,-0.28249 0.299724,-0.79926 v -0.34106 l -0.558106,0.0241 q -0.651123,0.0241 -0.950846,0.20671 -0.296278,0.18259 -0.296278,0.57533 0,0.29628 0.179145,0.45475 0.18259,0.15848 0.509874,0.15848 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path862" />
+      <path
+         d="m 48.427466,294.22715 q 0.120578,0 0.282498,-0.0241 0.161919,-0.0241 0.244602,-0.0551 v 0.47542 q -0.08613,0.0379 -0.268718,0.0723 -0.179145,0.0379 -0.36518,0.0379 -1.109321,0 -1.109321,-1.16788 v -2.20142 h -0.53399 v -0.29628 l 0.54088,-0.24804 0.248047,-0.80616 h 0.368625 v 0.86817 h 1.092096 v 0.48231 h -1.092096 v 2.18419 q 0,0.32729 0.15503,0.50299 0.158474,0.1757 0.437527,0.1757 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path864" />
+      <path
+         d="m 50.332607,294.66467 h -0.620118 v -3.78271 h 0.620118 z m -0.668349,-4.80591 q 0,-0.2067 0.103353,-0.29972 0.106798,-0.093 0.261827,-0.093 0.144694,0 0.251492,0.093 0.110243,0.093 0.110243,0.29972 0,0.20327 -0.110243,0.29973 -0.106798,0.0965 -0.251492,0.0965 -0.155029,0 -0.261827,-0.0965 -0.103353,-0.0965 -0.103353,-0.29973 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path866" />
+      <path
+         d="m 54.818119,292.76643 q 0,0.92673 -0.468533,1.44694 -0.468533,0.5202 -1.295356,0.5202 -0.51332,0 -0.909506,-0.23771 -0.396186,-0.24115 -0.609782,-0.68902 -0.213595,-0.44786 -0.213595,-1.04041 0,-0.92329 0.461642,-1.43661 0.461643,-0.51676 1.295356,-0.51676 0.799262,0 1.267795,0.5271 0.471979,0.52365 0.471979,1.42627 z m -2.855985,0 q 0,0.71313 0.279053,1.0852 0.282498,0.36863 0.830268,0.36863 0.54088,0 0.823378,-0.36863 0.282498,-0.37207 0.282498,-1.0852 0,-0.71314 -0.285943,-1.07143 -0.282498,-0.36173 -0.830268,-0.36173 -1.098986,0 -1.098986,1.43316 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path868" />
+      <path
+         d="m 58.380348,294.66467 v -2.41845 q 0,-0.46165 -0.206706,-0.68558 -0.20326,-0.22737 -0.640787,-0.22737 -0.582222,0 -0.847494,0.32383 -0.261827,0.3204 -0.261827,1.04731 v 1.96026 h -0.620117 v -3.78271 h 0.499538 l 0.09302,0.51676 h 0.03445 q 0.1757,-0.27905 0.492649,-0.43063 0.316948,-0.15503 0.699354,-0.15503 0.695909,0 1.033529,0.33762 0.337619,0.33761 0.337619,1.05075 v 2.46324 z"
+         style="font-style:normal;font-variant:normal;font-weight:normal;font-stretch:normal;font-size:7.05555582px;font-family:sans-serif;-inkscape-font-specification:'sans-serif, Normal';font-variant-ligatures:normal;font-variant-caps:normal;font-variant-numeric:normal;font-feature-settings:normal;text-align:start;writing-mode:lr-tb;text-anchor:start;fill:#f8f8f8;fill-opacity:1;stroke:#f8f8f8;stroke-width:0.26458335px;stroke-opacity:1"
+         id="path870" />
+    </g>
+  </g>
+</svg>

docs/archetypes/default.md

@@ -1,6 +0,0 @@
----
-title: "{{ replace .TranslationBaseName "-" " " | title }}"
-date: {{ .Date }}
-draft: true
----
-

docs/changelog.rst

@@ -0,0 +1,114 @@
+.. |break_config| replace:: **[CONFIG]**
+.. |break| replace:: **[BREAK]**
+.. |bugfix| replace:: [BUG]
+.. |docs| replace:: [DOCS]
+.. |feature| replace:: [FEATURE]
+
+.. _changelog:
+
+Changelog
+=========
+
+The changelog summarizes bugfixes that are deemed relevant for users and package maintainers.
+Developers should consult the git commit log or GitHub issue tracker.
+
+We use the following annotations for classifying changes:
+
+* |break_config| Change that breaks the config.
+  As a package maintainer, make sure to warn your users about config breakage somehow.
+* |break| Change that breaks interoperability or persistent state representation with previous releases.
+  As a package maintainer, make sure to warn your users about config breakage somehow.
+  Note that even updating the package on both sides might not be sufficient, e.g. if persistent state needs to be migrated to a new format.
+* |feature| Change that introduces new functionality.
+* |bugfix| Change that fixes a bug, no regressions or incompatibilities expected.
+* |docs| Change to the documentation.
+
+0.1
+---
+
+This release is a milestone for zrepl and required significant refactoring if not rewrites of substantial parts of the application.
+It breaks both configuration and transport format, and thus requires manual intervention and updates on both sides of a replication setup.
+
+.. DANGER::
+   The changes in the pruning system for this release require you to explicitly define **keep rules**:
+   for any snapshot that you want to keep, at least one rule must match.
+   This is different from previous releases where pruning only affected snapshots with the configured snapshotting prefix.
+   Make sure that snapshots to be kept or ignored by zrepl are covered, e.g. by using the ``regex`` keep rule.
+   :ref:`Learn more in the config docs... <prune>`
+
+
+Notes to Package Maintainers
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+* If the daemon crashes, the stack trace produced by the Go runtime and possibly diagnostic output of zrepl will be written to stderr.
+  This behavior is independent from the ``stdout`` outlet type.
+  Please make sure the stderr output of the daemon is captured somewhere.
+  To conserve precious stack traces, make sure that multiple service restarts do not directly discard previous stderr output.
+* Make it obvious for users how to set the ``GOTRACEBACK`` environment variable to ``GOTRACEBACK=crash``.
+  This functionality will cause SIGABRT on panics and can be used to capture a coredump of the panicking process.
+  To that extend, make sure that your package build system, your OS's coredump collection and the Go delve debugger work together.
+  Use your build system to package the Go program in `this tutorial on Go coredumps and the delve debugger <https://rakyll.org/coredumps/>`_ , and make sure the symbol resolution etc. work on coredumps captured from the binary produced by your build system. (Special focus on symbol stripping, etc.)
+* Use of ``ssh+stdinserver`` :ref:`transport <transport-ssh+stdinserver>` is no longer encouraged.
+  Please encourage users to use the new ``tcp`` or ``tls`` transports.
+  You might as well mention some of the :ref:`tunneling options listed here <transport-tcp-tunneling>`.
+
+Changes
+~~~~~~~
+
+* |feature| :issue:`55` : Push replication (see :ref:`push job <job-push>` and :ref:`sink job <job-sink>`)
+* |feature| :ref:`TCP Transport <transport-tcp>`
+* |feature| :ref:`TCP + TLS client authentication transport <transport-tcp+tlsclientauth>`
+* |feature| :issue:`78` :commit:`074f989` : Replication protocol rewrite
+
+  * Uses ``github.com/problame/go-streamrpc`` for RPC layer
+  * |break| Protocol breakage, update and restart of all zrepl daemons is required
+  * |feature| :issue:`83`:  Improved error handling of network-level errors (zrepl retries instead of failing the entire job)
+  * |bugfix| :issue:`75` :issue:`81`: use connection timeouts and protocol-level heartbeats
+  * |break| |break_config|: mappings are no longer supported
+
+    * Receiving sides (``pull`` and ``sink`` job) specify a single ``root_fs``.
+      Received filesystems are then stored *per client* in ``${root_fs}/${client_identity}``.
+
+* |feature| |break| |break_config| Manual snapshotting + triggering of replication
+
+  * |feature| :issue:`69`: include manually created snapshots in replication
+  * |break_config| ``manual`` and ``periodic`` :ref:`snapshotting types <job-snapshotting-spec>`
+  * |feature| ``zrepl signal wakeup JOB`` subcommand to trigger replication + pruning
+  * |feature| ``zrepl signal reset JOB`` subcommand to abort current replication + pruning
+
+* |feature| |break| |break_config| New pruning system
+
+  * The active side of a replication (pull or push) decides what to prune for both sender and receiver.
+    The RPC protocol is used to execute the destroy operations on the remote side.
+  * New pruning policies (see :ref:`configuration documentation <prune>` )
+
+    * The decision what snapshots shall be pruned is now made based on *keep rules*
+    * |feature| :issue:`68`: keep rule ``not_replicated`` prevents divergence of sender and receiver
+
+  * |feature| |break| Bookmark pruning is no longer necessary
+
+    * Per filesystem, zrepl creates a single bookmark (``#zrepl_replication_cursor``) and moves it forward with the most recent successfully replicated snapshot on the receiving side.
+    * Old bookmarks created by prior versions of zrepl (named like their corresponding snapshot) must be deleted manually.
+    * |break_config| ``keep_bookmarks`` parameter of the ``grid`` keep rule has been removed
+
+* |feature| ``zrepl status`` for live-updating replication progress (it's really cool!)
+* |feature| :issue:`67`: Expose `Prometheus <https://prometheus.io>`_ metrics via HTTP (:ref:`config docs <monitoring-prometheus>`)
+
+* |break_config| Logging outlet types must be specified using the ``type`` instead of ``outlet`` key
+* |break| :issue:`53`: CLI: ``zrepl control *`` subcommands have been made direct subcommands of ``zrepl *``
+* |bugfix| Goroutine leak on ssh transport connection timeouts
+* |bugfix| :issue:`81` :issue:`77` : handle failed accepts correctly (``source`` job)
+
+.. |lastrelease| replace:: 0.0.3
+
+Previous Releases
+-----------------
+
+.. NOTE::
+    Due to limitations in our documentation system, we only show the changelog since the last release and the time this documentation is built.
+    For the changelog of previous releases, use the version selection in the hosted version of these docs at `zrepl.github.io <https://zrepl.github.io>`_.
+    
+    
+
+
+W

docs/conf.py

@@ -0,0 +1 @@
+sphinxconf/conf.py

docs/config.toml

@@ -1,17 +0,0 @@
-baseURL = "https://zrepl.github.io"
-languageCode = "en-us"
-title = "zrepl - ZFS replication"
-theme = "docdock"
-
-[[menu.shortcuts]]
-pre = "<h3>More</h3>"
-name = "<i class='fa fa-github'></i> GitHub repo"
-identifier = "ds"
-url = "https://github.com/zrepl/zrepl"
-weight = 10
-
-[[menu.shortcuts]]
-name = "<i class='fa fa-twitter'></i> Primary Author on Twitter"
-url = "https://twitter.com/problame"
-weight = 11
-

docs/configuration.rst

@@ -0,0 +1,17 @@
+
+.. _configuration_toc:
+
+*************
+Configuration
+*************
+
+.. toctree::
+
+    configuration/preface
+    configuration/jobs
+    configuration/transports
+    configuration/filter_syntax
+    configuration/prune
+    configuration/logging
+    configuration/monitoring
+    configuration/misc

docs/configuration/filter_syntax.rst

@@ -0,0 +1,67 @@
+.. include:: ../global.rst.inc
+
+.. _pattern-filter:
+
+Filter Syntax
+=============
+
+For :ref:`source jobs <job-source>` and :ref:`push jobs <job-push>`, a filesystem filter must be defined (field ``filesystems``).
+A filter takes a filesystem path (in the ZFS filesystem hierarchy) as parameter and returns ``true`` (pass) or ``false`` (block).
+
+A filter is specified as a **YAML dictionary** with patterns as keys and booleans as values.
+The following rules determine which result is chosen for a given filesystem path:
+
+* More specific path patterns win over less specific ones
+* Non-wildcard patterns (full path patterns) win over *subtree wildcards* (`<` at end of pattern)
+* If the path in question does not match any pattern, the result is ``false``.
+
+The **subtree wildcard** ``<`` means "the dataset left of ``<`` and all its children".
+   
+.. TIP::
+  You can try out patterns for a configured job using the ``zrepl test filesystems`` subcommand for push and source jobs.
+
+Examples
+--------
+
+Full Access
+~~~~~~~~~~~
+
+The following configuration will allow access to all filesystems.
+
+::
+
+   jobs:
+   - type: source
+     filesystems: {
+       "<": true,
+     }
+     ...
+
+
+Fine-grained
+~~~~~~~~~~~~
+
+The following configuration demonstrates all rules presented above.
+
+::
+
+    jobs:
+    - type: source
+      filesystems: {
+        "tank<": true,          # rule 1
+        "tank/foo<": false,     # rule 2
+        "tank/foo/bar": true,  # rule 3
+      }
+      ...
+
+
+Which rule applies to given path, and what is the result?
+
+::
+
+    tank/foo/bar/loo => 2    false
+    tank/bar         => 1    true
+    tank/foo/bar     => 3    true
+    zroot            => NONE false
+    tank/var/log     => 1    true
+

docs/configuration/jobs.rst

@@ -0,0 +1,267 @@
+.. include:: ../global.rst.inc
+
+.. |serve-transport| replace:: :ref:`serve specification<transport>`
+.. |connect-transport| replace:: :ref:`connect specification<transport>`
+.. |snapshotting-spec| replace:: :ref:`snapshotting specification <job-snapshotting-spec>`
+.. |pruning-spec| replace:: :ref:`pruning specification <prune>`
+.. |filter-spec| replace:: :ref:`filter specification<pattern-filter>`
+
+.. _job:
+
+Job Types & Replication
+=======================
+
+Overview & Terminology
+----------------------
+
+A *job* is the unit of activity tracked by the zrepl daemon and configured in the |mainconfig|.
+Every job has a unique ``name``, a ``type`` and type-dependent fields which are documented on this page.
+
+Replication always happens between a pair of jobs: one is the **active side**, and one the **passive side**.
+The active side executes the replication logic whereas the passive side responds to requests after checking the active side's permissions.
+For communication, the active side connects to the passive side using a :ref:`transport <transport>` and starts issuing remote procedure calls (RPCs).
+
+The following table shows how different job types can be combined to achieve both push and pull mode setups:
+
++-----------------------+--------------+----------------------------------+-----------------------------------------------+
+| Setup name            | active side  | passive side                     | use case                                      |
++=======================+==============+==================================+===============================================+
+| Push mode             | ``push``     | ``sink``                         | * Laptop backup                               |
+|                       |              |                                  | * NAS behind NAT to offsite                   |
++-----------------------+--------------+----------------------------------+-----------------------------------------------+
+| Pull mode             | ``pull``     | ``source``                       | * Central backup-server for many nodes        |
+|                       |              |                                  | * Remote server to NAS behind NAT             |
++-----------------------+--------------+----------------------------------+-----------------------------------------------+
+| Local replication     | | ``push`` + ``sink`` in one config             | * Backup FreeBSD boot pool                    |
+|                       | | with :ref:`local transport <transport-local>` |                                               |
++-----------------------+--------------+----------------------------------+-----------------------------------------------+
+
+How the Active Side Works
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The active side (:ref:`push <job-push>` and :ref:`pull <job-pull>` job) executes the replication and pruning logic:
+
+* Wakeup because of finished snapshotting (``push`` job) or pull interval ticker (``pull`` job).
+* Connect to the corresponding passive side using a :ref:`transport <transport>` and instantiate an RPC client.
+* Replicate data from the sending to the receiving side.
+* Prune on sender & receiver.
+
+.. TIP::
+  The progress of the active side can be watched live using the ``zrepl status`` subcommand.
+
+How the Passive Side Works
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The passive side (:ref:`sink <job-sink>` and :ref:`source <job-source>`) waits for connections from the corresponding active side,
+using the transport listener type specified in the ``serve`` field of the job configuration.
+Each transport listener provides a client's identity to the passive side job.
+It uses the client identity for access control:
+
+* The ``sink`` job only allows pushes to those ZFS filesystems to the active side that are located below ``root_fs/${client_identity}``.
+* The ``source`` job has a whitelist of client identities that are allowed pull access.
+
+.. TIP::
+   The implementation of the ``sink`` job requires that the connecting client identities be a valid ZFS filesystem name components.
+
+How Replication Works
+~~~~~~~~~~~~~~~~~~~~~
+
+One of the major design goals of the replication module is to avoid any duplication of the nontrivial logic.
+As such, the code works on abstract senders and receiver **endpoints**, where typically one will be implemented by a local program object and the other is an RPC client instance.
+Regardless of push- or pull-style setup, the logic executes on the active side, i.e. in the ``push`` or ``pull`` job.
+
+The following steps take place during replication and can be monitored using the ``zrepl status`` subcommand:
+
+* Plan the replication:
+
+  * Compare sender and receiver filesystem snapshots
+  * Build the **replication plan**
+
+    * Per filesystem, compute a diff between sender and receiver snapshots
+    * Build a list of replication steps
+
+      * If possible, use incremental sends (``zfs send -i``)
+      * Otherwise, use full send of most recent snapshot on sender
+      * Give up on filesystems that cannot be replicated without data loss
+
+  * Retry on errors that are likely temporary (i.e. network failures).
+  * Give up on filesystems where a permanent error was received over RPC.
+
+* Execute the plan
+
+  * Perform replication steps in the following order:
+    Among all filesystems with pending replication steps, pick the filesystem whose next replication step's snapshot is the oldest.
+  * After a successful replication step, update the replication cursor bookmark (see below)
+   
+The idea behind the execution order of replication steps is that if the sender snapshots all filesystems simultaneously at fixed intervals, the receiver will have all filesystems snapshotted at time ``T1`` before the first snapshot at ``T2 = T1 + $interval`` is replicated.
+
+.. _replication-cursor-bookmark:
+
+The **replication cursor bookmark** ``#zrepl_replication_cursor`` is kept per filesystem on the sending side of a replication setup:
+It is a bookmark of the most recent successfully replicated snapshot to the receiving side.
+It is is used by the :ref:`not_replicated <prune-keep-not-replicated>` keep rule to identify all snapshots that have not yet been replicated to the receiving side.
+Regardless of whether that keep rule is used, the bookmark ensures that replication can always continue incrementally.
+
+.. ATTENTION::
+
+    Currently, zrepl does not replicate filesystem properties.
+    Whe receiving a filesystem, it is never mounted (`-u` flag)  and `mountpoint=none` is set.
+    This is temporary and being worked on :issue:`24`.
+
+
+.. _job-snapshotting-spec:
+
+Taking Snaphots
+---------------
+
+The ``push`` and ``source`` jobs can automatically take periodic snapshots of the filesystems matched by the ``filesystems`` filter field.
+The snapshot names are composed of a user-defined prefix followed by a UTC date formatted like ``20060102_150405_000``.
+We use UTC because it will avoid name conflicts when switching time zones or between summer and winter time.
+
+For ``push`` jobs, replication is automatically triggered after all filesystems have been snapshotted.
+
+::
+
+    jobs:
+    - type: push
+      filesystems: {
+        "<": true,
+        "tmp": false
+      }
+      snapshotting:
+        type: periodic
+        prefix: zrepl_
+        interval: 10m
+      ...
+
+There is also a ``manual`` snapshotting type, which covers the following use cases:
+
+* Existing infrastructure for automatic snapshots: you only want to use zrepl for replication.
+* Run scripts before and after taking snapshots (like locking database tables).
+  We are working on better integration for this use case: see :issue:`74`.
+
+Note that you will have to trigger replication manually using the ``zrepl signal wakeup JOB`` subcommand in that case.
+
+::
+
+   jobs:
+   - type: push
+     filesystems: {
+       "<": true,
+       "tmp": false
+     }
+     snapshotting:
+       type: manual
+     ...
+
+
+.. _job-push:
+
+Job Type ``push``
+-----------------
+
+.. list-table::
+    :widths: 20 80
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - = ``push``
+    * - ``name``
+      - unique name of the job
+    * - ``connect``
+      - |connect-transport|
+    * - ``filesystems``
+      - |filter-spec| for filesystems to be snapshotted and pushed to the sink
+    * - ``snapshotting``
+      - |snapshotting-spec|
+    * - ``pruning``
+      - |pruning-spec|
+
+Example config: :sampleconf:`/push.yml`
+
+.. _job-sink:
+
+Job Type ``sink``
+-----------------
+
+.. list-table::
+    :widths: 20 80
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - = ``sink``
+    * - ``name``
+      - unique name of the job
+    * - ``serve``
+      - |serve-transport|
+    * - ``root_fs``
+      - ZFS dataset path are received to
+        ``$root_fs/$client_identity``
+
+Example config: :sampleconf:`/sink.yml`
+
+.. _job-pull:
+
+Job Type ``pull``
+-----------------
+
+.. list-table::
+    :widths: 20 80
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - = ``pull``
+    * - ``name``
+      - unique name of the job
+    * - ``connect``
+      - |connect-transport|
+    * - ``root_fs``
+      - ZFS dataset path are received to
+        ``$root_fs/$client_identity``
+    * - ``interval``
+      - Interval at which to pull from the source job
+    * - ``pruning``
+      - |pruning-spec|
+
+Example config: :sampleconf:`/pull.yml`
+
+.. _job-source:
+
+Job Type ``source``
+-------------------
+
+.. list-table::
+    :widths: 20 80
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - = ``source``
+    * - ``name``
+      - unique name of the job
+    * - ``serve``
+      - |serve-transport|
+    * - ``filesystems``
+      - |filter-spec| for filesystems to be snapshotted and exposed to connecting clients
+    * - ``snapshotting``
+      - |snapshotting-spec|
+
+Example config: :sampleconf:`/source.yml`
+
+.. _replication-local:
+
+Local replication
+-----------------
+
+If you have the need for local replication (most likely between two local storage pools), you can use the :ref:`local transport type <transport-local>` to connect a local push job to a local sink job.
+
+Example config: :sampleconf:`/local.yml`.
+
+

docs/configuration/logging.rst

@@ -0,0 +1,212 @@
+.. include:: ../global.rst.inc
+
+.. _logging:
+
+Logging
+=======
+
+zrepl uses structured logging to provide users with easily processable log messages.
+
+Logging outlets are configured in the ``global`` section of the |mainconfig|.
+
+::
+
+    global:
+      logging:
+    
+        - type: OUTLET_TYPE
+          level: MINIMUM_LEVEL
+          format: FORMAT
+    
+        - type: OUTLET_TYPE
+          level: MINIMUM_LEVEL
+          format: FORMAT
+    
+        ...
+    
+    jobs: ...
+
+.. _logging-error-outlet:
+
+.. ATTENTION::
+    The **first outlet is special**: if an error writing to any outlet occurs, the first outlet receives the error and can print it.
+    Thus, the first outlet must be the one that always works and does not block, e.g. ``stdout``, which is the default.
+
+.. _logging-default-config:
+
+Default Configuration
+---------------------
+
+By default, the following logging configuration is used
+
+::
+
+    global:
+      logging:
+    
+        - type: "stdout"
+          level:  "warn"
+          format: "human"
+
+Building Blocks
+---------------
+
+The following sections document the semantics of the different log levels, formats and outlet types.
+
+.. _logging-levels:
+
+Levels
+~~~~~~
+
+.. list-table::
+    :widths: 10 10 80
+    :header-rows: 1
+
+    * - Level
+      - SHORT
+      - Description
+    * - ``error``
+      - ``ERRO``
+      - immediate action required
+    * - ``warn``
+      - ``WARN``
+      - symptoms for misconfiguration, soon expected failure, etc.
+    * - ``info``
+      - ``INFO``
+      - explains what happens without too much detail
+    * - ``debug``
+      - ``DEBG``
+      - tracing information, state dumps, etc. useful for debugging.
+
+Incorrectly classified messages are considered a bug and should be reported.
+
+.. _logging-formats:
+
+Formats
+~~~~~~~
+
+.. list-table::
+    :widths: 10 90
+    :header-rows: 1
+
+    * - Format
+      - Description
+    * - ``human``
+      - prints job and subsystem into brackets before the actual message,
+        followed by remaining fields in logfmt style
+    * - ``logfmt``
+      - `logfmt <https://brandur.org/logfmt>`_ output. zrepl uses `this Go package <https://github.com/go-logfmt/logfmt>`_.
+    * - ``json``
+      - JSON formatted output. Each line is a valid JSON document. Fields are marshaled by
+        ``encoding/json.Marshal()``, which is particularly useful for processing in
+        log aggregation or when processing state dumps.
+
+Outlets
+~~~~~~~
+
+Outlets are the destination for log entries.
+
+.. _logging-outlet-stdout:
+
+``stdout`` Outlet
+-----------------
+
+.. list-table::
+    :widths: 10 90
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - ``stdout``
+    * - ``level``
+      -  minimum  :ref:`log level <logging-levels>`
+    * - ``format``
+      - output :ref:`format <logging-formats>`
+    * - ``time``
+      - always include time in output (``true`` or ``false``)
+    * - ``color``
+      - colorize output according to log level (``true`` or ``false``)
+
+Writes all log entries with minimum level ``level`` formatted by ``format`` to stdout.
+If stdout is a tty, interactive usage is assumed and both ``time`` and ``color`` are set to ``true``.
+
+Can only be specified once.
+
+``syslog`` Outlet
+-----------------
+.. list-table::
+    :widths: 10 90
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - ``syslog``
+    * - ``level``
+      -  minimum  :ref:`log level <logging-levels>`
+    * - ``format``
+      - output :ref:`format <logging-formats>`
+    * - ``retry_interval``
+      - Interval between reconnection attempts to syslog (default = 0)
+
+Writes all log entries formatted by ``format`` to syslog.
+On normal setups, you should not need to change the ``retry_interval``.
+
+Can only be specified once.
+
+``tcp`` Outlet
+--------------
+
+.. list-table::
+    :widths: 10 90
+    :header-rows: 1
+
+    * - Parameter
+      - Comment
+    * - ``type``
+      - ``tcp``
+    * - ``level``
+      -  minimum  :ref:`log level <logging-levels>`
+    * - ``format``
+      - output :ref:`format <logging-formats>`
+    * - ``net``
+      - ``tcp`` in most cases
+    * - ``address``
+      - remote network, e.g. ``logs.example.com:10202``
+    * - ``retry_interval``
+      - Interval between reconnection attempts to ``address``
+    * - ``tls``
+      - TLS config (see below)
+
+Establishes a TCP connection to ``address`` and sends log messages with minimum level ``level`` formatted by ``format``.
+If ``tls`` is not specified, an unencrypted connection is established.
+If ``tls`` is specified, the TCP connection is secured with TLS + Client Authentication.
+The latter is particularly useful in combination with log aggregation services.
+
+.. list-table::
+    :widths: 10 90
+    :header-rows: 1
+
+    * - Parameter
+      - Description
+    * - ``ca``
+      - PEM-encoded certificate authority that signed the remote server's TLS certificate
+    * - ``cert``
+      - PEM-encoded client certificate identifying this zrepl daemon toward the remote server
+    * - ``key``
+      - PEM-encoded, unencrypted client private key identifying this zrepl daemon toward the remote server
+
+.. WARNING::
+
+    zrepl drops log messages to the TCP outlet if the underlying connection is not fast enough.
+    Note that TCP buffering in the kernel must first run full becfore messages are dropped.
+
+    Make sure to always configure a ``stdout`` outlet as the special error outlet to be informed about problems
+    with the TCP outlet (see :ref:`above <logging-error-outlet>` ).
+
+
+.. NOTE::
+
+    zrepl uses Go's ``crypto/tls`` and ``crypto/x509`` packages and leaves all but the required fields in ``tls.Config`` at their default values.
+    In case of a security defect in these packages, zrepl has to be rebuilt because Go binaries are statically linked.

docs/configuration/misc.rst

@@ -0,0 +1,62 @@
+Miscellaneous
+=============
+
+.. _conf-runtime-directories:
+
+Runtime Directories & UNIX Sockets
+----------------------------------
+
+zrepl daemon creates various UNIX sockets to allow communicating with it:
+
+* the :ref:`transport-ssh+stdinserver` transport connects to a socket named after ``client_identity`` parameter
+* the ``control`` CLI subcommand connects to a defined control socket
+
+There is no further authentication on these sockets.
+Therefore we have to make sure they can only be created and accessed by ``zrepl daemon``.
+In fact, ``zrepl daemon`` will not bind a socket to a path in a directory that is world-accessible.
+
+The directories can be configured in the main configuration file, the defaults are provided below:
+
+::
+
+    global:
+      control:
+        sockpath: /var/run/zrepl/control
+      serve:
+        stdinserver:
+          sockdir: /var/run/zrepl/stdinserver
+
+
+Durations & Intervals
+---------------------
+
+Interval & duration fields in job definitions, pruning configurations, etc. must match the following regex:
+
+::
+
+    var durationStringRegex *regexp.Regexp = regexp.MustCompile(`^\s*(\d+)\s*(s|m|h|d|w)\s*$`)
+    // s = second, m = minute, h = hour, d = day, w = week (7 days)
+
+Super-Verbose Job Debugging
+---------------------------
+
+You have probably landed here because you opened an issue on GitHub and some developer told you to do this...
+So just read the annotated comments ;)
+
+::
+
+    job:
+    - name: ...
+      ...
+     # JOB DEBUGGING OPTIONS
+      # should be equal for all job types, but each job implements the debugging itself
+      debug:
+        conn: # debug the io.ReadWriteCloser connection
+          read_dump: /tmp/connlog_read   # dump results of Read() invocations to this file
+          write_dump: /tmp/connlog_write # dump results of Write() invocations to this file
+        rpc: # debug the RPC protocol implementation
+          log: true # log output from rpc layer to the job log
+
+.. ATTENTION::
+
+    Connection dumps will almost certainly contain your or other's private data. Do not share it in a bug report.

docs/configuration/monitoring.rst

@@ -0,0 +1,29 @@
+.. include:: ../global.rst.inc
+
+.. _monitoring:
+
+Monitoring
+==========
+
+Monitoring endpoints are configured in the ``global.monitoring`` section of the |mainconfig|.
+
+.. _monitoring-prometheus:
+
+Prometheus
+----------
+
+zrepl can expose `Prometheus metrics <https://prometheus.io/docs/instrumenting/exposition_formats/>`_ via HTTP.
+The ``listen`` attribute is a `net.Listen <https://golang.org/pkg/net/#Listen>`_  string for tcp, e.g. ``:9091`` or ``127.0.0.1:9091``.
+
+The Prometheues monitoring job appears in the ``zrepl control`` job list and may be specified **at most once**.
+At the time of writing, there is no stability guarantee on the exported metrics.
+
+::
+
+    global:
+      monitoring:
+        - type: prometheus
+          listen: ':9091'
+
+
+

docs/configuration/preface.rst

@@ -0,0 +1,39 @@
+
+.. _configuration_preface:
+
+=======
+Preface
+=======
+
+-----------------------
+Configuration File Path
+-----------------------
+
+zrepl searches for its main configuration file in the following locations (in that order):
+
+* If set, the location specified via the global ``--config`` flag
+* ``/etc/zrepl/zrepl.yml``
+* ``/usr/local/etc/zrepl/zrepl.yml``
+
+The examples in the :ref:`tutorial` or the :sampleconf:`/` directory should provide a good starting point.
+
+-------------------
+Runtime Directories
+-------------------
+
+zrepl requires runtime directories for various UNIX sockets --- they are documented in the :ref:`config file<conf-runtime-directories>`.
+Your package maintainer / init script should take care of creating them.
+Alternatively, for default settings, the following should to the trick.
+
+::
+
+    mkdir -p /var/run/zrepl/stdinserver
+    chmod -R 0700 /var/run/zrepl
+
+
+----------
+Validating
+----------
+
+The config can be validated using the ``zrepl configcheck`` subcommand.
+

docs/configuration/prune.rst

@@ -0,0 +1,157 @@
+.. _prune:
+
+Pruning Policies
+================
+
+In zrepl, *pruning* means *destroying snapshots*.
+Pruning must happen on both sides of a replication or the systems would inevitable run out of disk space at some point.
+
+Typically, the requirements to temporal resolution and maximum retention time differ per side.
+For example, when using zrepl to back up a busy database server, you will want high temporal resolution (snapshots every 10 min) for the last 24h in case of administrative disasters, but cannot afford to store them for much longer because you might have high turnover volume in the database.
+On the receiving side, you may have more disk space available, or need to comply with other backup retention policies.
+
+zrepl uses a set of  **keep rules** to determine which snapshots shall be kept per filesystem.
+**A snapshot that is not kept by any rule is destroyed.**
+The keep rules are **evaluated on the active side** (:ref:`push <job-push>` or :ref:`pull job <job-pull>`) of the replication setup, for both active and passive side, after replication completed or was determined to have failed permanently.
+
+Example Configuration:
+
+::
+
+   jobs:
+     - type: push
+       name: ...
+       connect: ...
+       filesystems: {
+         "<": true,
+         "tmp": false
+       }
+       snapshotting:
+         type: periodic
+         prefix: zrepl_
+         interval: 10m
+       pruning:
+         keep_sender:
+           - type: not_replicated
+           # make sure manually created snapshots by the administrator are kept
+           - type: regex
+             regex: "^manual_.*"
+           - type: grid
+             grid: 1x1h(keep=all) | 24x1h | 14x1d
+             regex: "^zrepl_.*"
+         keep_receiver:
+           - type: grid
+             grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+             regex: "^zrepl_.*"
+           # manually created snapshots will be kept forever on receiver
+
+.. DANGER::
+    You might have **existing snapshots** of filesystems affected by pruning which you want to keep, i.e. not be destroyed by zrepl.
+    Make sure to actually add the necessary ``regex`` keep rules on both sides, like with ``manual`` in the example above.
+
+.. ATTENTION::
+
+    It is currently not possible to define pruning on a source job.
+    The source job creates snapshots, which means that extended replication downtime will fill up the source's zpool with snapshots, since pruning is directed by the corresponding active side (pull job).
+    If this is a potential risk for you, consider using :ref:`push mode <job-push>`.
+
+
+.. _prune-keep-not-replicated:
+
+Policy ``not_replicated``
+-------------------------
+
+::
+
+   jobs:
+   - type: push
+     pruning:
+       keep_sender:
+       - type: not_replicated
+     ...
+
+``not_replicated`` keeps all snapshots that have not been replicated to the receiving side.
+It only makes sense to specify this rule on a sender (source or push job).
+The state required to evaluate this rule is stored in the :ref:`replication cursor bookmark <replication-cursor-bookmark>` on the sending side.
+
+.. _prune-keep-retention-grid:
+
+Policy ``grid``
+---------------
+
+::
+
+    jobs:
+    - type: pull
+      pruning:
+        keep_receiver:
+        - type: grid
+          regex: "^zrepl_.*"
+          grid: 1x1h(keep=all) | 24x1h | 35x1d | 6x30d
+                │                │
+                └─ one hour interval
+                                 │
+                                 └─ 24 adjacent one-hour intervals
+      ...
+
+The retention grid can be thought of as a time-based sieve:
+The ``grid`` field specifies a list of adjacent time intervals:
+the left edge of the leftmost (first) interval is the ``creation`` date of the youngest snapshot.
+All intervals to its right describe time intervals further in the past.
+
+Each interval carries a maximum number of snapshots to keep.
+It is specified via ``(keep=N)``, where ``N`` is either ``all`` (all snapshots are kept) or a positive integer.
+The default value is **keep=1**.
+
+The following procedure happens during pruning:
+
+#. The list of snapshots is filtered by the regular expression in ``regex``.
+   Only snapshots names that match the regex are considered for this rule, all others are not affected.
+#. The filtered list of snapshots is sorted by ``creation``
+#. The left edge of the first interval is aligned to the ``creation`` date of the youngest snapshot
+#. A list of buckets is created, one for each interval
+#. The list of snapshots is split up into the buckets.
+#. For each bucket
+
+   #. the contained snapshot list is sorted by creation.
+   #. snapshots from the list, oldest first, are destroyed until the specified ``keep`` count is reached.
+   #. all remaining snapshots on the list are kept.
+
+
+.. _prune-keep-last-n:
+
+Policy ``last_n``
+-----------------
+
+::
+
+   jobs:
+     - type: push
+       pruning:
+         keep_receiver:
+         - type: last_n
+           count: 10
+     ...
+
+``last_n`` keeps the last ``count`` snapshots (last = youngest = most recent creation date).
+
+.. _prune-keep-regex:
+
+Policy ``regex``
+----------------
+
+::
+
+   jobs:
+     - type: push
+       pruning:
+         keep_receiver:
+         - type: regex
+           regex: "^(zrepl|manual)_.*"
+    ...
+
+``regex`` keeps all snapshots whose names are matched by the regular expressionin ``regex``.
+Like all other regular expression fields in prune policies, zrepl uses Go's `regexp.Regexp <https://golang.org/pkg/regexp/#Compile>`_ Perl-compatible regular expressions (`Syntax <https://golang.org/pkg/regexp/syntax>`_).
+
+
+

docs/configuration/transports.rst

@@ -0,0 +1,340 @@
+.. highlight:: bash
+
+.. _transport:
+
+Transports
+==========
+
+The zrepl RPC layer uses **transports** to establish a single, bidirectional data stream between an active and passive job.
+On the passive (serving) side, the transport also provides the **client identity** to the upper layers:
+this string is used for access control and separation of filesystem sub-trees in :ref:`sink jobs <job-sink>`.
+Transports are specified in the ``connect`` or ``serve`` section of a job definition.
+
+.. contents::
+
+.. ATTENTION::
+
+    The **client identities must be valid ZFS dataset path components**
+    because the :ref:`sink job <job-sink>` uses ``${root_fs}/${client_identity}`` to determine the client's subtree.
+
+.. _transport-tcp:
+
+``tcp`` Transport
+-----------------
+
+The ``tcp`` transport uses plain TCP, which means that the data is **not encrypted** on the wire.
+Clients are identified by their IPv4 or IPv6 addresses, and the client identity is established through a mapping on the server.
+
+This transport may also be used in conjunction with network-layer encryption and/or VPN tunnels to provide encryption on the wire.
+To make the IP-based client authentication effective, such solutions should provide authenticated IP addresses.
+Some options to consider:
+
+.. _transport-tcp-tunneling:
+
+* `WireGuard <https://www.wireguard.com/>`_: Linux-focussed, in-kernel TLS
+* `OpenVPN <https://openvpn.net/>`_: Cross-platform VPN, uses tun on \*nix
+* `IPSec <https://en.wikipedia.org/wiki/IPsec>`_: Properly standardized, in-kernel network-layer VPN
+* `spiped <http://www.tarsnap.com/spiped.html>`_: think of it as an encrypted pipe between two servers
+* SSH
+
+  * `sshuttle <https://sshuttle.readthedocs.io/en/stable/overview.html>`_: VPN-like solution, but using SSH
+  * `SSH port forwarding <https://help.ubuntu.com/community/SSH/OpenSSH/PortForwarding>`_: Systemd user unit & make it start before the zrepl service.
+
+Serve
+~~~~~
+
+::
+
+    jobs:
+    - type: sink
+      serve:
+        type: tcp
+        listen: ":8888"
+        clients: {
+          "192.168.122.123" : "mysql01"
+          "192.168.122.123" : "mx01"
+        }
+      ...
+
+Connect
+~~~~~~~
+
+::
+
+    jobs:
+     - type: push
+       connect:
+         type: tcp
+         address: "10.23.42.23:8888"
+         dial_timeout: # optional, default 10s
+       ...
+
+.. _transport-tcp+tlsclientauth:
+
+``tls`` Transport
+-----------------
+
+The ``tls`` transport uses TCP + TLS with client authentication using client certificates.
+The client identity is the common name (CN) presented in the client certificate.
+It is recommended to set up a dedicated CA infrastructure for this transport, e.g. using OpenVPN's `EasyRSA <https://github.com/OpenVPN/easy-rsa>`_.
+For a simple 2-machine setup, see the :ref:`instructions below<transport-tcp+tlsclientauth-2machineopenssl>`.
+
+The implementation uses `Go's TLS library <https://golang.org/pkg/crypto/tls/>`_.
+Since Go binaries are statically linked, you or your distribution need to recompile zrepl when vulnerabilities in that library are disclosed.
+
+All file paths are resolved relative to the zrepl daemon's working directory.
+Specify absolute paths if you are unsure what directory that is (or find out from your init system).
+
+Serve
+~~~~~
+
+::
+
+    jobs:
+      - type: sink
+        root_fs: "pool2/backup_laptops"
+        serve:
+          type: tls
+          listen: ":8888"
+          ca: /etc/zrepl/ca.crt
+          cert: /etc/zrepl/prod.crt
+          key: /etc/zrepl/prod.key
+          client_cns:
+            - "laptop1"
+            - "homeserver"
+
+The ``ca`` field specified the certificate authority used to validate client certificates.
+The ``client_cns`` list specifies a list of accepted client common names (which are also the client identities for this transport).
+
+Connect
+~~~~~~~
+
+::
+
+    jobs:
+    - type: pull
+      connect:
+        type: tls
+        address: "server1.foo.bar:8888"
+        ca: /etc/zrepl/ca.crt
+        cert: /etc/zrepl/backupserver.crt
+        key:  /etc/zrepl/backupserver.key
+        server_cn: "server1"
+        dial_timeout: # optional, default 10s
+
+The ``ca`` field specifies the CA which signed the server's certificate (``serve.cert``).
+The ``server_cn`` specifies the expected common name (CN) of the server's certificate.
+It overrides the hostname specified in ``address``.
+The connection fails if either do not match.
+
+.. _transport-tcp+tlsclientauth-2machineopenssl:
+
+Self-Signed Certificates
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+Tools like `EasyRSA <https://github.com/OpenVPN/easy-rsa>`_ make it easy to manage CA infrastructure for multiple clients, e.g. a central zrepl backup server (in sink mode).
+However, for a two-machine setup, self-signed certificates distributed using an out-of-band mechanism will also work just fine:
+
+Suppose you have a push-mode setup, with `backups.example.com` running the :ref:`sink job <job-sink>`, and `prod.example.com` running the :ref:`push job <job-push>`.
+Run the following OpenSSL commands on each host, substituting HOSTNAME in both filenames and the interactive input prompt by OpenSSL:
+
+.. code-block:: bash
+   :emphasize-lines: 1-5,24
+
+   openssl req -x509 -sha256 -nodes \
+      -newkey rsa:4096 \
+      -days 365 \
+      -keyout HOSTNAME.key \
+      -out HOSTNAME.crt
+
+   #Generating a 4096 bit RSA private key
+   #................++++
+   #.++++
+   #writing new private key to 'backups.key'
+   #-----
+   #You are about to be asked to enter information that will be incorporated
+   #into your certificate request.
+   #What you are about to enter is what is called a Distinguished Name or a DN.
+   #There are quite a few fields but you can leave some blank
+   #For some fields there will be a default value,
+   #If you enter '.', the field will be left blank.
+   #-----
+   #Country Name (2 letter code) [XX]:
+   #State or Province Name (full name) []:
+   #Locality Name (eg, city) [Default City]:
+   #Organization Name (eg, company) [Default Company Ltd]:
+   #Organizational Unit Name (eg, section) []:
+   #Common Name (eg, your name or your server's hostname) []:HOSTNAME
+   #Email Address []:
+
+Now copy each machine's ``HOSTNAME.crt`` to the other machine's ``/etc/zrepl/HOSTNAME.crt``, for example using `scp`.
+The serve & connect configuration will thus look like the following:
+
+::
+
+   # on backups.example.com
+   - type: sink
+     serve:
+       type: tls
+       listen: ":8888"
+       ca: "/etc/zrepl/prod.example.com.crt"
+       cert: "/etc/zrepl/backups.example.com.crt"
+       key: "/etc/zrepl/backups.example.com.key"
+       client_cns:
+         - "prod.example.com"
+     ...
+
+   # on prod.example.com
+   - type: push
+     connect:
+       type: tls
+       address:"backups.example.com:8888"
+       ca: /etc/zrepl/backups.example.com.crt
+       cert: /etc/zrepl/prod.example.com.crt
+       key:  /etc/zrepl/prod.example.com.key
+       server_cn: "backups.example.com"
+     ...
+
+
+.. _transport-ssh+stdinserver:
+
+``ssh+stdinserver`` Transport
+-----------------------------
+
+``ssh+stdinserver`` is inspired by `git shell <https://git-scm.com/docs/git-shell>`_ and `Borg Backup <https://borgbackup.readthedocs.io/en/stable/deployment.html>`_.
+It is provided by the Go package ``github.com/problame/go-netssh``.
+
+.. ATTENTION::
+
+   ``ssh+stdinserver`` has inferior error detection and handling compared to the ``tcp`` and ``tls`` transports.
+   If you require tested timeout & retry handling, use ``tcp`` or ``tls`` transports, or help improve package go-netssh.
+
+.. _transport-ssh+stdinserver-serve:
+
+Serve
+~~~~~
+
+::
+
+    jobs:
+    - type: source
+      serve:
+        type: stdinserver
+        client_identities:
+        - "client1"
+        - "client2"
+      ...
+
+First of all, note that ``type=stdinserver`` in this case:
+Currently, only ``connect.type=ssh+stdinserver`` can connect to a ``serve.type=stdinserver``, but we want to keep that option open for future extensions.
+
+The serving job opens a UNIX socket named after ``client_identity`` in the runtime directory.
+In our example above, that is ``/var/run/zrepl/stdinserver/client1`` and ``/var/run/zrepl/stdinserver/client2``.
+
+On the same machine, the ``zrepl stdinserver $client_identity`` command connects to ``/var/run/zrepl/stdinserver/$client_identity``.
+It then passes its stdin and stdout file descriptors to the zrepl daemon via *cmsg(3)*.
+zrepl daemon in turn combines them into an object implementing ``net.Conn``:
+a ``Write()`` turns into a write to stdout, a ``Read()`` turns into a read from stdin.
+
+Interactive use of the ``stdinserver`` subcommand does not make much sense.
+However, we can force its execution when a user with a particular SSH pubkey connects via SSH.
+This can be achieved with an entry in the ``authorized_keys`` file of the serving zrepl daemon.
+
+::
+
+    # for OpenSSH >= 7.2
+    command="zrepl stdinserver CLIENT_IDENTITY",restrict CLIENT_SSH_KEY
+    # for older OpenSSH versions
+    command="zrepl stdinserver CLIENT_IDENTITY",no-port-forwarding,no-X11-forwarding,no-pty,no-agent-forwarding,no-user-rc CLIENT_SSH_KEY
+
+* CLIENT_IDENTITY is substituted with an entry from ``client_identities`` in our example
+* CLIENT_SSH_KEY is substituted with the public part of the SSH keypair specified in the ``connect.identity_file`` directive on the connecting host.
+
+.. NOTE::
+
+    You may need to adjust the ``PermitRootLogin`` option in ``/etc/ssh/sshd_config`` to ``forced-commands-only`` or higher for this to work.
+    Refer to sshd_config(5) for details.
+
+To recap, this is of how client authentication works with the ``ssh+stdinserver`` transport:
+
+* Connections to the ``/var/run/zrepl/stdinserver/${client_identity}`` UNIX socket are blindly trusted by zrepl daemon.
+  The connection client identity is the name of the socket, i.e. ``${client_identity}``.
+* Thus, the runtime directory must be private to the zrepl user (this is checked by zrepl daemon)
+* The admin of the host with the serving zrepl daemon controls the ``authorized_keys`` file.
+* Thus, the administrator controls the mapping ``PUBKEY -> CLIENT_IDENTITY``.
+
+.. _transport-ssh+stdinserver-connect:
+
+Connect
+~~~~~~~
+
+::
+
+    jobs:
+    - type: pull
+      connect:
+        type: ssh+stdinserver
+        host: prod.example.com
+        user: root
+        port: 22
+        identity_file: /etc/zrepl/ssh/identity
+        options: # optional, default [], `-o` arguments passed to ssh
+        - "Compression=on"
+        dial_timeout: # optional, default 10s, max time.Duration until initial handshake is completed
+
+The connecting zrepl daemon
+
+#. Creates a pipe
+#. Forks
+#. In the forked process
+
+   #. Replaces forked stdin and stdout with the corresponding pipe ends
+   #. Executes the ``ssh`` binary found in ``$PATH``.
+
+      #. The identity file (``-i``) is set to ``$identity_file``.
+      #. The remote user, host and port correspond to those configured.
+      #. Further options can be specified using the ``options`` field, which appends each entry in the list to the command line using ``-o $entry``.
+
+#. Wraps the pipe ends in a ``net.Conn`` and returns it to the RPC layer.
+
+As discussed in the section above, the connecting zrepl daemon expects that ``zrepl stdinserver $client_identity`` is  executed automatically via an ``authorized_keys`` file entry.
+
+The ``known_hosts`` file used by the ssh command must contain an entry for ``connect.host`` prior to starting zrepl.
+Thus, run the following on the pulling host's command line (substituting ``connect.host``):
+
+::
+
+    ssh -i /etc/zrepl/ssh/identity root@prod.example.com
+
+.. NOTE::
+
+    The environment variables of the underlying SSH process are cleared. ``$SSH_AUTH_SOCK`` will not be available.
+    It is suggested to create a separate, unencrypted SSH key solely for that purpose.
+
+
+.. _transport-local:
+
+``local`` Transport
+-------------------
+
+The local transport can be used to implement :ref:`local replication <replication-local>`, i.e., push replication between a push and sink job defined in the same configuration file.
+
+The ``listener_name`` is analogous to a hostname and must match between ``serve`` and ``connect``.
+The ``client_identity`` is used by the sink as documented above.
+
+::
+
+    jobs:
+    - type: sink
+      serve:
+        type: local
+        listener_name: localsink
+      ...
+
+    - type: push
+      connect:
+        type: local
+        listener_name: localsink
+        client_identity: local_backup
+      ...
+

docs/content/_header.md

@@ -1 +0,0 @@
-zrepl Documentation

docs/content/_index.md

@@ -1,45 +0,0 @@
-+++
-title = "zrepl - ZFS replication"
-+++
-
-# zrepl - ZFS replication
-
-`zrepl` is a tool for replicating ZFS filesystems.
-
-{{% panel theme="danger" header="Important" %}}
-`zrepl` as well as this documentation is still under active
-development. Some of the features below are not implemented yet. Use & test at your own risk ;)
-{{% / panel  %}}
-
-## Main Features
-
-* filesystem replication
-    * local & over network (SSH)
-    * push & pull mode
-    * snapshots & bookmarks support
-    * feature-negotiation for
-        * resumable `send & receive`
-        * compressed `send & receive`
-        * raw encrypted `send & receive` (as soon as it is available)
-    * access control checks when pulling datasets
-    * [flexible mappings]({{< ref "configuration/overview.md#mapping-filter-syntax" >}}) for filesystems
-* automatic snapshot creation
-    * periodic interval
-* automatic snapshot pruning
-    * [Retention Grid]({{< ref "configuration/snapshots.md#retention-grid" >}})
-   
-## Contributing
-
-`zrepl` is usable but nowhere near a stable release right now -  we are happy
-about contributors!
-
-* Explore the codebase
-    * These docs live in the `docs/` subdirectory
-* Document non-obvious / confusing / plain broken things you encounter when using `zrepl` for the first time
-* Check the *Issues* and *Projects* sections for things to do ;)
-
-{{% panel header="<i class='fa fa-github'></i> Getting your code merged"%}}
-[The <i class='fa fa-github'></i> GitHub repository](https://github.com/zrepl/zrepl) is where all development happens.
-Open your issue / PR there.
-{{% /panel %}}
-

docs/content/configuration/_index.md

@@ -1,6 +0,0 @@
-+++
-title = "Configuration"
-alwaysopen = true
-+++
-
-{{% children description="true" %}}