KAFKA-18615: StreamThread *-ratio metrics suffer from sampling bias #21160
+87
−14
Conversation
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
mjsax
reviewed
Dec 16, 2025
streams/src/main/java/org/apache/kafka/streams/processor/internals/StreamThread.java
Outdated
Show resolved
Hide resolved
mjsax
reviewed
Dec 16, 2025
Member
mjsax
left a comment
mjsax
left a comment
There was a problem hiding this comment.
Should we also update the description of each metric, saying it's windowed (eg in ThreadMetrics.java)
bbejeck
reviewed
Dec 16, 2025
Member
bbejeck
left a comment
bbejeck
left a comment
There was a problem hiding this comment.
Thanks @aliehsaeedii - overall LGTM - can we add a description to docs/upgrade.html section for 4.2?
mjsax
reviewed
Dec 17, 2025
Member
mjsax
left a comment
mjsax
left a comment
There was a problem hiding this comment.
Overall LGTM. Few more minor things.
| <li>The behavior of <code>org.apache.kafka.streams.KafkaStreams#removeStreamThread</code> has been changed. The consumer has no longer remove once <code>removeStreamThread</code> finished. | ||
| Instead, consumer would be kicked off from the group after <code>org.apache.kafka.streams.processor.internals.StreamThread</code> completes its <code>run</code> function. | ||
| </li> | ||
| <li>TThe streams thread metrics <code>commit-ratio</code>, <code>process-ratio</code>, <code>punctuate-ratio</code>, and <code>poll-ratio</code> have been updated. |
Member
There was a problem hiding this comment.
Suggested change
| <li>TThe streams thread metrics <code>commit-ratio</code>, <code>process-ratio</code>, <code>punctuate-ratio</code>, and <code>poll-ratio</code> have been updated. | |
| <li>The streams thread metrics <code>commit-ratio</code>, <code>process-ratio</code>, <code>punctuate-ratio</code>, and <code>poll-ratio</code> have been updated. |
Member
There was a problem hiding this comment.
@bbejeck Did you intent docs/upgrade.html? Not sure if this change it too small for top level upgrade notes and if docs/streams/upgrade-guide.html would be better suited?
| boolean cleanRun = false; | ||
| try { | ||
| taskManager.init(); | ||
| initLatencyWindowsIfNeeded(System.currentTimeMillis()); |
Member
There was a problem hiding this comment.
We should never call System.curentTimeMillis()directly, but use time.milliseconds() instead.
| pollLatencyWindowedSum.record(metricsConfig, 0.0, now); | ||
| this.totalCommitLatencyWindowedSum.record(metricsConfig, 0, now); | ||
| this.processLatencyWindowedSum.record(metricsConfig, 0, now); | ||
| this.punctuateLatencyWindowedSum.record(metricsConfig, 0, now); |
| this.totalCommitLatencyWindowedSum.record(metricsConfig, totalCommitLatency, now); | ||
| this.processLatencyWindowedSum.record(metricsConfig, processLatency, now); | ||
| this.punctuateLatencyWindowedSum.record(metricsConfig, punctuateLatency, now); | ||
| this.runOnceLatencyWindowedSum.record(metricsConfig, runOnceLatency, now); |
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.
This PR implements KAFKA-18615 by adding a windowSum aggregation that
computes the sum of values over a time window, so that
commit-ratio,poll-ratio,process-ratio, andpunctuate-ratiorepresent the ratioof the
{action}over a window duration rather than a single iteration.The effective window duration is whatever you configure for metrics:
metrics.sample.window.ms(per-sample window length)times metrics.num.samples(number of rolling windows)With the default Kafka metrics config, that is typically:
metrics.sample.window.ms = 30000 msmetrics.num.samples = 2→ ~60seconds total rolling window.