prevent duplicate targets after OTel Collector Prometheus relabeling#4381
Merged
swiatekm merged 4 commits intoopen-telemetry:mainfrom Oct 7, 2025
Merged
Conversation
Contributor
Author
|
Hi @swiatekm @jaronoff97 This is the bugfix-related PR that I've split out from the original changes. During testing, I noticed that the unit test exhibits flaky behavior - it sometimes passes and sometimes fails. I suspect this is due to race conditions caused by asynchronous event processing and cache synchronization delays.If this flaky test issue persists during your review, would it be appropriate to include a fix for this unit test in this PR as well? |
Contributor
If you are able to figure out why it fails, I'd love a fix! |
Contributor
Author
|
Hello @swiatekm Evidence: |
Contributor
|
@mike9421 Feel free to include that timeout increase in this PR, then, and we'll see if it helps universally. |
swiatekm
reviewed
Sep 26, 2025
Contributor
swiatekm
left a comment
swiatekm
left a comment
There was a problem hiding this comment.
This largely looks good to me. What I'd like to understand as well is the performance impact. Can you run this benchmark first on main, and then on this PR, and post the benchstat comparison results?
Contributor
Author
Hi @swiatekm I've run benchmark tests before and after the changes to compare performance impact: Before changes: a7855fa8.txt Summary:
Benchmark output:
|
Contributor
|
I did 10 runs of both main and this branch, here's the benchstat: So we have around a 15% increase in CPU usage and memory for targets with relabeling enabled. This is a fair amount, but considering this is fixing a bug and we're going to improve the overall efficiency of the whole system by doing target relabeling in the target allocator in a subsequent change, I think this is acceptable. WDYT @jaronoff97 @pavolloffay ? |
swiatekm
approved these changes
Sep 27, 2025
|
This issue is rendering Target Allocator unusable for us as well, could you please review this MR or propose a workaround in the meantime? @swiatekm |
Contributor
There is no workaround other than not rewriting the address label. The bug only happens as a result of that. For this PR, I've already approved it, but it has a performance impact, so I'd like opinions from my fellow maintainers before merging it. The earliest it might be released is two weeks from now, in |
jaronoff97
approved these changes
Oct 7, 2025
Contributor
jaronoff97
left a comment
jaronoff97
left a comment
There was a problem hiding this comment.
I think the perf hit is worth the actual functionality fix. We've done a fair amount of work in making the TA more efficient, so I think a minor regression is acceptable.
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
4 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.
Description:
Extract hash calculation fixes into a separate PR
Link to tracking Issue(s):
Testing:
Documentation: