daemon/active: implement watchdog to handle stuck replication / pruners

ActiveSide.do() can only run sequentially, i.e. we cannot run
replication and pruning in parallel. Why?

* go-streamrpc only allows one active request at a time
(this is bad design and should be fixed at some point)
* replication and pruning are implemented independently, but work on the
same resources (snapshots)

A: pruning might destroy a snapshot that is planned to be replicated
B: replication might replicate snapshots that should be pruned

We do not have any resource management / locking for A and B, but we
have a use case where users don't want their machine fill up with
snapshots if replication does not work.
That means we _have_ to run the pruners.

A further complication is that we cannot just cancel the replication
context after a timeout and move on to the pruner: it could be initial
replication and we don't know how long it will take.
(And we don't have resumable send & recv yet).

With the previous commits, we can implement the watchdog using context
cancellation.
Note that the 'MadeProgress()' calls can only be placed right before
non-error state transition. Otherwise, we could end up in a live-lock.
This commit is contained in:
Christian Schwarz
2018-10-19 16:27:05 +02:00
parent 4ede99b08c
commit 69bfcb7bed
8 changed files with 240 additions and 35 deletions
+20 -1
View File
@@ -5,6 +5,7 @@ import (
"net"
"os"
"sync/atomic"
"time"
)
type NetConnLogger struct {
@@ -101,6 +102,14 @@ func (c *ChainedReader) Read(buf []byte) (n int, err error) {
type ByteCounterReader struct {
reader io.ReadCloser
// called & accessed synchronously during Read, no external access
cb func(full int64)
cbEvery time.Duration
lastCbAt time.Time
bytesSinceLastCb int64
// set atomically because it may be read by multiple threads
bytes int64
}
@@ -110,13 +119,23 @@ func NewByteCounterReader(reader io.ReadCloser) *ByteCounterReader {
}
}
func (b *ByteCounterReader) SetCallback(every time.Duration, cb func(full int64)) {
b.cbEvery = every
b.cb = cb
}
func (b *ByteCounterReader) Close() error {
return b.reader.Close()
}
func (b *ByteCounterReader) Read(p []byte) (n int, err error) {
n, err = b.reader.Read(p)
atomic.AddInt64(&b.bytes, int64(n))
full := atomic.AddInt64(&b.bytes, int64(n))
now := time.Now()
if b.cb != nil && now.Sub(b.lastCbAt) > b.cbEvery {
b.cb(full)
b.lastCbAt = now
}
return n, err
}