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

Don't fail writes due to full WAL disk #3136

Merged
merged 3 commits into from
Jan 7, 2021
Merged

Conversation

owen-d
Copy link
Member

@owen-d owen-d commented Jan 7, 2021

No description provided.

@codecov-io
Copy link

codecov-io commented Jan 7, 2021

Codecov Report

Merging #3136 (c1cb2e7) into master (fcabfec) will increase coverage by 0.02%.
The diff coverage is 95.23%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master    #3136      +/-   ##
==========================================
+ Coverage   63.05%   63.08%   +0.02%     
==========================================
  Files         188      188              
  Lines       16210    16239      +29     
==========================================
+ Hits        10221    10244      +23     
- Misses       5049     5060      +11     
+ Partials      940      935       -5     
Impacted Files Coverage Δ
pkg/ingester/instance.go 61.59% <92.00%> (+3.16%) ⬆️
pkg/ingester/ingester.go 48.61% <100.00%> (+0.47%) ⬆️
pkg/ingester/metrics.go 100.00% <100.00%> (ø)
pkg/promtail/positions/positions.go 46.80% <0.00%> (-11.71%) ⬇️
pkg/promtail/targets/file/filetarget.go 66.43% <0.00%> (+2.09%) ⬆️
pkg/querier/queryrange/limits.go 95.83% <0.00%> (+4.16%) ⬆️


1) No space left on disk

In the event the underlying WAL disk is full, Loki will not fail incoming writes, but neither will it log them to the WAL. In this case, the persistence guarantees across process restarts will not hold.
Copy link
Collaborator

@slim-bean slim-bean Jan 7, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Interesting thought, feel free to tell me this is too much scope.

If we know writing to the WAL is failing can we force a flush on shutdown?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this is a great idea. Otherwise, we'd end up having to remove an ingester from traffic and wait for chunk_idle to elapse before shutting it down.

@@ -161,8 +163,13 @@ func (i *instance) Push(ctx context.Context, req *logproto.PushRequest) error {

if !record.IsEmpty() {
if err := i.wal.Log(record); err != nil {
return err
if e, ok := err.(*os.PathError); ok && e.Err == syscall.ENOSPC {
i.metrics.walDiskFullFailures.Inc()
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It would be nice if we could log that this was happening, but we don't want to spam the logs.

Thoughts on using a boolean kept in the instance to log something one time, and then perhaps clear the bool and log that the error is cleared if writes start succeeding again?

@pull-request-size pull-request-size bot added size/L and removed size/M labels Jan 7, 2021
Copy link
Collaborator

@slim-bean slim-bean left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants