Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add an exponential backoff to TCP listeners to avoid fast loops in error scenarios #11588

Merged
merged 3 commits into from May 12, 2021
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
3 changes: 3 additions & 0 deletions changelog/11588.txt
@@ -0,0 +1,3 @@
```release-note:improvement
core: Add a small (<1s) exponential backoff to failed TCP listener Accept failures.
```
24 changes: 24 additions & 0 deletions vault/cluster/cluster.go
Expand Up @@ -279,6 +279,15 @@ func (cl *Listener) Run(ctx context.Context) error {
close(closeCh)
}()

// baseDelay is the initial delay after an Accept() error before attempting again
const baseDelay = 5 * time.Millisecond

// maxDelay is the maximum delay after an Accept() error before attempting again.
// In the case that this function is error-looping, it will delay the shutdown check.
// Therefore, changes to maxDelay may have an effect on the latency of shutdown.
const maxDelay = 1 * time.Second

var loopDelay time.Duration
for {
if atomic.LoadUint32(cl.shutdown) > 0 {
return
Expand All @@ -298,8 +307,23 @@ func (cl *Listener) Run(ctx context.Context) error {
if conn != nil {
conn.Close()
}

if loopDelay == 0 {
loopDelay = baseDelay
} else {
loopDelay *= 2
}

if loopDelay > maxDelay {
loopDelay = maxDelay
}

time.Sleep(loopDelay)
continue
}
// No error, reset loop delay
loopDelay = 0

if conn == nil {
continue
}
Expand Down