suryakantade opened a new pull request, #22621:
URL: https://github.com/apache/kafka/pull/22621
Summary
Fixes the flaky testBalancePartitionLeaders test that was failing ~3% of CI
runs.
The root cause is a timing issue: the leader imbalance check interval (1s)
combined with a 10× timeout multiplier left only a 10s window for the periodic
electPreferred task to fire and rebalance partitions. Under CI
thread-scheduling jitter (KafkaEventQueue uses real wall-clock
cond.awaitNanos()), this was insufficient.
Changes
- Reduce leaderImbalanceCheckIntervalNs from 1s to 100ms — the periodic
task fires more frequently, reducing sensitivity to thread-scheduling delays.
- Increase the waitForCondition timeout multiplier from 10× to 100× (still
10s total) — gives ample headroom for slow CI environments.
- Add Javadoc explaining the test lifecycle and why the low check interval
is necessary.
Jira: KAFKA-19092 (https://issues.apache.org/jira/browse/KAFKA-19092)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
