test(kafkajs): drop retries:0 from header-disable producer tests#9106
Merged
Conversation
The second sendBatch/send in these tests is a real broker call that the
test expects to succeed after the stub is restored. A fresh topic's first
produce routinely returns the retryable NOT_LEADER_FOR_PARTITION while
metadata propagates; kafkajs normally refreshes metadata and retries it,
but retry:{retries:0} stripped that safety net, surfacing the transient
error as a hard KafkaJSNonRetriableError and flaking CI.
retries:0 bought nothing for the first call it was meant to speed up: the
stubbed UNKNOWN error is non-retryable, so it already fails on the first
attempt regardless of the retry count. Removing it restores the retry on
the real call while leaving the stubbed-rejection assertions unchanged.
Contributor
Overall package sizeSelf size: 6.39 MB Dependency sizes| name | version | self size | total size | |------|---------|-----------|------------| | import-in-the-middle | 3.2.0 | 104.26 kB | 843.44 kB | | opentracing | 0.14.7 | 194.81 kB | 194.81 kB | | dc-polyfill | 0.1.11 | 25.74 kB | 25.74 kB |🤖 This report was automatically generated by heaviest-objects-in-the-universe |
🎉 All green!🧪 All tests passed 🎯 Code Coverage (details) 🔗 Commit SHA: 9a870f4 | Docs | Datadog PR Page | Give us feedback! |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #9106 +/- ##
==========================================
- Coverage 93.69% 93.67% -0.02%
==========================================
Files 889 889
Lines 50856 50856
Branches 11830 11830
==========================================
- Hits 47647 47638 -9
- Misses 3209 3218 +9 Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
BenchmarksBenchmark execution time: 2026-06-27 15:03:56 Comparing candidate commit 9a870f4 in PR branch Found 0 performance improvements and 0 performance regressions! Performance is the same for 2254 metrics, 32 unstable metrics.
|
pabloerhard
approved these changes
Jun 30, 2026
dd-octo-sts Bot
pushed a commit
that referenced
this pull request
Jul 1, 2026
The second sendBatch/send in these tests is a real broker call that the
test expects to succeed after the stub is restored. A fresh topic's first
produce routinely returns the retryable NOT_LEADER_FOR_PARTITION while
metadata propagates; kafkajs normally refreshes metadata and retries it,
but retry:{retries:0} stripped that safety net, surfacing the transient
error as a hard KafkaJSNonRetriableError and flaking CI.
retries:0 bought nothing for the first call it was meant to speed up: the
stubbed UNKNOWN error is non-retryable, so it already fails on the first
attempt regardless of the retry count. Removing it restores the retry on
the real call while leaving the stubbed-rejection assertions unchanged.
rochdev
pushed a commit
that referenced
this pull request
Jul 2, 2026
The second sendBatch/send in these tests is a real broker call that the
test expects to succeed after the stub is restored. A fresh topic's first
produce routinely returns the retryable NOT_LEADER_FOR_PARTITION while
metadata propagates; kafkajs normally refreshes metadata and retries it,
but retry:{retries:0} stripped that safety net, surfacing the transient
error as a hard KafkaJSNonRetriableError and flaking CI.
retries:0 bought nothing for the first call it was meant to speed up: the
stubbed UNKNOWN error is non-retryable, so it already fails on the first
attempt regardless of the retry count. Removing it restores the retry on
the real call while leaving the stubbed-rejection assertions unchanged.
rochdev
pushed a commit
that referenced
this pull request
Jul 2, 2026
The second sendBatch/send in these tests is a real broker call that the
test expects to succeed after the stub is restored. A fresh topic's first
produce routinely returns the retryable NOT_LEADER_FOR_PARTITION while
metadata propagates; kafkajs normally refreshes metadata and retries it,
but retry:{retries:0} stripped that safety net, surfacing the transient
error as a hard KafkaJSNonRetriableError and flaking CI.
retries:0 bought nothing for the first call it was meant to speed up: the
stubbed UNKNOWN error is non-retryable, so it already fails on the first
attempt regardless of the retry count. Removing it restores the retry on
the real call while leaving the stubbed-rejection assertions unchanged.
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
2 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.
Summary
The
when broker rejects headers with UNKNOWN_SERVER_ERRORproducer tests (sendandsendBatchvariants) issue a real broker call as their second send, expecting it to succeed once the stub is restored. A fresh topic's first produce routinely returns the retryableNOT_LEADER_FOR_PARTITIONwhile metadata propagates — kafkajs normally refreshes metadata and retries it, butretry: { retries: 0 }stripped that safety net, surfacing the transient error as a hardKafkaJSNonRetriableErrorand flaking thekafkajs (latest)job.retries: 0bought nothing for the first (stubbed) call it was meant to speed up: the stubbedUNKNOWNerror is non-retryable, so it already rejects on the first attempt regardless of the retry count (verified against a local 2.2.4 broker — first call fails in ~5 ms, propagates raw, plugin still seestype: 'UNKNOWN'). The 6–8 s seen on 1.4.0 is that version's inherent error-path slowness, already covered bythis.timeout(30000), not the retry count.Why
Removing
retries: 0restores kafkajs's own metadata-refresh-and-retry on the real send while leaving the stubbed-rejection assertions and the header-disable behaviour unchanged. Forcing one transientNOT_LEADER_FOR_PARTITIONon the real send against a live broker: withretries: 0it rejects (KafkaJSNumberOfRetriesExceededcaused byNOT_LEADER_FOR_PARTITION, reproducing the CI failure); with default retries it resolves witherrorCode: 0.Test plan
kafkajsmatrix (oldest + latest) green