Improve consumer pending count tracking during stream contention#6297
Merged
derekcollison merged 1 commit intomainfrom Dec 23, 2024
Merged
Improve consumer pending count tracking during stream contention#6297derekcollison merged 1 commit intomainfrom
derekcollison merged 1 commit intomainfrom
Conversation
Signed-off-by: Maurice van Veen <github@mauricevanveen.com>
kozlovic
reviewed
Dec 23, 2024
| // Either we have not reached the message yet, or we've hit the race condition | ||
| // when there is contention at the beginning of the stream. In which case we can | ||
| // only decrement if the ack floor is still low enough to be able to detect it. | ||
| if o.isFilteredMatch(subj) && sseq > o.asflr && (sseq >= o.sseq || !wasPending) { |
Member
There was a problem hiding this comment.
nit: given that o.isFilteredMatch(subj) can be more expensive than the other checks, I would have put it last:
if sseq > o.asflr && (sseq >= o.sseq || !wasPending) && o.isFilteredMatch(subj) {
derekcollison
added a commit
that referenced
this pull request
Dec 24, 2024
Makes the change suggested in #6297 (comment) Signed-off-by: Maurice van Veen <github@mauricevanveen.com>
derekcollison
added a commit
that referenced
this pull request
Jan 8, 2025
Reverts the change made in #6297. The change did (somewhat) improve the reliability of the drifting tests if the `o.npc--` was done as a result of contention, but resulted in a regression if `o.npc--` was done twice when a message was acked that did not move the ack floor up for WorkQueue/Interest retention. This PR fixes what would otherwise have been a regression. We really should try to fix the race condition itself though (outside of this PR). Without fixing it the pending count can be incorrect with no way to be resolved unless all messages are consumed, or we'd need to manually recalculate. Signed-off-by: Maurice van Veen <github@mauricevanveen.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
The drifting tests would occasionally fail due to the consumer pending count drifting. This was due to a race condition described on
checkNumPending:This PR doesn't fix this race condition, but improves the tracking which improves test reliability. If the race condition happens we can still check if the deleted message is between
o.asflrand theo.sseqthat's skipped ahead. In which case we can still decrement the pending count (o.npc) if the message is not pending/delivered. This improves the reliability of the pending count tracking as long as the ack floor hasn't moved up yet.Signed-off-by: Maurice van Veen github@mauricevanveen.com