[#307] add package trace, integrate it with logging, and adopt it throughout zrepl

package trace:

- introduce the concept of tasks and spans, tracked as linked list within ctx
    - see package-level docs for an overview of the concepts
    - **main feature 1**: unique stack of task and span IDs
        - makes it easy to follow a series of log entries in concurrent code
    - **main feature 2**: ability to produce a chrome://tracing-compatible trace file
        - either via an env variable or a `zrepl pprof` subcommand
        - this is not a CPU profile, we already have go pprof for that
        - but it is very useful to visually inspect where the
          replication / snapshotter / pruner spends its time
          ( fixes #307 )

usage in package daemon/logging:

- goal: every log entry should have a trace field with the ID stack from package trace

- make `logging.GetLogger(ctx, Subsys)` the authoritative `logger.Logger` factory function
    - the context carries a linked list of injected fields which
      `logging.GetLogger` adds to the logger it returns
    - `logging.GetLogger` also uses package `trace` to get the
      task-and-span-stack and injects it into the returned logger's fields
This commit is contained in:
Christian Schwarz
2020-04-11 15:49:41 +02:00
parent bcb5965617
commit 10a14a8c50
75 changed files with 1934 additions and 462 deletions
+2 -1
View File
@@ -1,6 +1,7 @@
package client
import (
"context"
"encoding/json"
"fmt"
"os"
@@ -29,7 +30,7 @@ var ConfigcheckCmd = &cli.Subcommand{
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 {
Run: func(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
formatMap := map[string]func(interface{}){
"": func(i interface{}) {},
"pretty": func(i interface{}) {
+2 -5
View File
@@ -48,14 +48,13 @@ var migratePlaceholder0_1Args struct {
dryRun bool
}
func doMigratePlaceholder0_1(sc *cli.Subcommand, args []string) error {
func doMigratePlaceholder0_1(ctx context.Context, sc *cli.Subcommand, args []string) error {
if len(args) != 0 {
return fmt.Errorf("migration does not take arguments, got %v", args)
}
cfg := sc.Config()
ctx := context.Background()
allFSS, err := zfs.ZFSListMapping(ctx, zfs.NoFilter())
if err != nil {
return errors.Wrap(err, "cannot list filesystems")
@@ -124,7 +123,7 @@ var fail = color.New(color.FgRed)
var migrateReplicationCursorSkipSentinel = fmt.Errorf("skipping this filesystem")
func doMigrateReplicationCursor(sc *cli.Subcommand, args []string) error {
func doMigrateReplicationCursor(ctx context.Context, sc *cli.Subcommand, args []string) error {
if len(args) != 0 {
return fmt.Errorf("migration does not take arguments, got %v", args)
}
@@ -137,8 +136,6 @@ func doMigrateReplicationCursor(sc *cli.Subcommand, args []string) error {
return fmt.Errorf("exiting migration after error")
}
ctx := context.Background()
v1cursorJobs := make([]job.Job, 0, len(cfg.Jobs))
for i, j := range cfg.Jobs {
if jobs[i].Name() != j.Name() {
+4 -61
View File
@@ -1,67 +1,10 @@
package client
import (
"errors"
"log"
"os"
"github.com/zrepl/zrepl/cli"
"github.com/zrepl/zrepl/config"
"github.com/zrepl/zrepl/daemon"
)
var pprofArgs struct {
daemon.PprofServerControlMsg
}
import "github.com/zrepl/zrepl/cli"
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")
Use: "pprof",
SetupSubcommands: func() []*cli.Subcommand {
return []*cli.Subcommand{PprofListenCmd, pprofActivityTraceCmd}
},
}
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")
}
+44
View File
@@ -0,0 +1,44 @@
package client
import (
"context"
"io"
"log"
"os"
"golang.org/x/net/websocket"
"github.com/zrepl/zrepl/cli"
)
var pprofActivityTraceCmd = &cli.Subcommand{
Use: "activity-trace ZREPL_PPROF_HOST:ZREPL_PPROF_PORT",
Short: "attach to zrepl daemon with activated pprof listener and dump an activity-trace to stdout",
Run: runPProfActivityTrace,
}
func runPProfActivityTrace(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
log := log.New(os.Stderr, "", 0)
die := func() {
log.Printf("exiting after error")
os.Exit(1)
}
if len(args) != 1 {
log.Printf("exactly one positional argument is required")
die()
}
url := "ws://" + args[0] + "/debug/zrepl/activity-trace" // FIXME dont' repeat that
log.Printf("attaching to activity trace stream %s", url)
ws, err := websocket.Dial(url, "", url)
if err != nil {
log.Printf("error: %s", err)
die()
}
_, err = io.Copy(os.Stdout, ws)
return err
}
+68
View File
@@ -0,0 +1,68 @@
package client
import (
"context"
"errors"
"log"
"os"
"github.com/zrepl/zrepl/cli"
"github.com/zrepl/zrepl/config"
"github.com/zrepl/zrepl/daemon"
)
var pprofListenCmd struct {
daemon.PprofServerControlMsg
}
var PprofListenCmd = &cli.Subcommand{
Use: "listen off | [on TCP_LISTEN_ADDRESS]",
Short: "start a http server exposing go-tool-compatible profiling endpoints at TCP_LISTEN_ADDRESS",
Run: func(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
if len(args) < 1 {
goto enargs
}
switch args[0] {
case "on":
pprofListenCmd.Run = true
if len(args) != 2 {
return errors.New("must specify TCP_LISTEN_ADDRESS as second positional argument")
}
pprofListenCmd.HttpListenAddress = args[1]
case "off":
if len(args) != 1 {
goto enargs
}
pprofListenCmd.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, pprofListenCmd.PprofServerControlMsg, struct{}{})
if err != nil {
log.Printf("error sending control message: %s", err)
die()
}
log.Printf("finished")
}
+3 -1
View File
@@ -1,6 +1,8 @@
package client
import (
"context"
"github.com/pkg/errors"
"github.com/zrepl/zrepl/cli"
@@ -11,7 +13,7 @@ import (
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 {
Run: func(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
return runSignalCmd(subcommand.Config(), args)
},
}
+2 -1
View File
@@ -1,6 +1,7 @@
package client
import (
"context"
"fmt"
"io"
"math"
@@ -180,7 +181,7 @@ var StatusCmd = &cli.Subcommand{
Run: runStatus,
}
func runStatus(s *cli.Subcommand, args []string) error {
func runStatus(ctx context.Context, s *cli.Subcommand, args []string) error {
httpc, err := controlHttpClient(s.Config().Global.Control.SockPath)
if err != nil {
return err
+1 -1
View File
@@ -17,7 +17,7 @@ import (
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 {
Run: func(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
return runStdinserver(subcommand.Config(), args)
},
}
+4 -6
View File
@@ -39,7 +39,7 @@ var testFilter = &cli.Subcommand{
Run: runTestFilterCmd,
}
func runTestFilterCmd(subcommand *cli.Subcommand, args []string) error {
func runTestFilterCmd(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
if testFilterArgs.job == "" {
return fmt.Errorf("must specify --job flag")
@@ -49,7 +49,6 @@ func runTestFilterCmd(subcommand *cli.Subcommand, args []string) error {
}
conf := subcommand.Config()
ctx := context.Background()
var confFilter config.FilesystemsFilter
job, err := conf.Job(testFilterArgs.job)
@@ -137,10 +136,9 @@ var testPlaceholder = &cli.Subcommand{
Run: runTestPlaceholder,
}
func runTestPlaceholder(subcommand *cli.Subcommand, args []string) error {
func runTestPlaceholder(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
var checkDPs []*zfs.DatasetPath
ctx := context.Background()
// all actions first
if testPlaceholderArgs.all {
@@ -197,11 +195,11 @@ var testDecodeResumeToken = &cli.Subcommand{
Run: runTestDecodeResumeTokenCmd,
}
func runTestDecodeResumeTokenCmd(subcommand *cli.Subcommand, args []string) error {
func runTestDecodeResumeTokenCmd(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
if testDecodeResumeTokenArgs.token == "" {
return fmt.Errorf("token argument must be specified")
}
token, err := zfs.ParseResumeToken(context.Background(), testDecodeResumeTokenArgs.token)
token, err := zfs.ParseResumeToken(ctx, testDecodeResumeTokenArgs.token)
if err != nil {
return err
}
+2 -1
View File
@@ -1,6 +1,7 @@
package client
import (
"context"
"fmt"
"os"
@@ -25,7 +26,7 @@ var VersionCmd = &cli.Subcommand{
SetupFlags: func(f *pflag.FlagSet) {
f.StringVar(&versionArgs.Show, "show", "", "version info to show (client|daemon)")
},
Run: func(subcommand *cli.Subcommand, args []string) error {
Run: func(ctx context.Context, subcommand *cli.Subcommand, args []string) error {
versionArgs.Config = subcommand.Config()
versionArgs.ConfigErr = subcommand.ConfigParsingError()
return runVersionCmd()
+1 -3
View File
@@ -28,7 +28,7 @@ var zabsCmdCreateStepHold = &cli.Subcommand{
},
}
func doZabsCreateStep(sc *cli.Subcommand, args []string) error {
func doZabsCreateStep(ctx context.Context, sc *cli.Subcommand, args []string) error {
if len(args) > 0 {
return errors.New("subcommand takes no arguments")
}
@@ -44,8 +44,6 @@ func doZabsCreateStep(sc *cli.Subcommand, args []string) error {
return errors.Errorf("jobid must be set")
}
ctx := context.Background()
v, err := zfs.ZFSGetFilesystemVersion(ctx, f.target)
if err != nil {
return errors.Wrapf(err, "get info about target %q", f.target)
+1 -2
View File
@@ -32,9 +32,8 @@ var zabsCmdList = &cli.Subcommand{
},
}
func doZabsList(sc *cli.Subcommand, args []string) error {
func doZabsList(ctx context.Context, sc *cli.Subcommand, args []string) error {
var err error
ctx := context.Background()
if len(args) > 0 {
return errors.New("this subcommand takes no positional arguments")
+2 -4
View File
@@ -43,9 +43,8 @@ var zabsCmdReleaseStale = &cli.Subcommand{
SetupFlags: registerZabsReleaseFlags,
}
func doZabsReleaseAll(sc *cli.Subcommand, args []string) error {
func doZabsReleaseAll(ctx context.Context, sc *cli.Subcommand, args []string) error {
var err error
ctx := context.Background()
if len(args) > 0 {
return errors.New("this subcommand takes no positional arguments")
@@ -68,10 +67,9 @@ func doZabsReleaseAll(sc *cli.Subcommand, args []string) error {
return doZabsRelease_Common(ctx, abstractions)
}
func doZabsReleaseStale(sc *cli.Subcommand, args []string) error {
func doZabsReleaseStale(ctx context.Context, sc *cli.Subcommand, args []string) error {
var err error
ctx := context.Background()
if len(args) > 0 {
return errors.New("this subcommand takes no positional arguments")