fix(test): add missing unlock in TestPersistLFDiscardStats #1924
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
TestPersistLFDiscardStats
is currently flaky. E.g., 1, 2, 3 (but this goes back further). To reproduce the problem:From the stack traces in the logs, this looks like a deadlock. The test currently sleeps for 2 seconds to allow for compaction to complete. Assuming compaction does not complete in this time, and we make the call to
db.vlog.discardStats.Lock()
, seems to be enough to put us into this deadlock. This is supported experimentally; running the above test command with this change no longer produces the deadlock.There is another sleep in the test to give discardStats a chance to be populated by the newly generated discardStats. We have not seen this issue in CI, but when running this test at very high frequencies (i.e. with
-count=160
), we start to see this failure as well. This is mitigated by increasing the sleep.