unknowntpo opened a new pull request, #23781: URL: https://github.com/apache/kafka/pull/23781
`ShareConsumerPerformance` used `--timeout` as a limit for the whole run, although it is documented as the maximum time between returned records. A run longer than the timeout was interrupted, the tool printed only the header line, and the benchmark failed with `ValueError: could not convert string to float`. With only that fixed, the benchmarks could still pass with 0 records read: the first share group in a cluster has to wait for `__share_group_state` to be created, which can take longer than the timeout. So the benchmarks now read and release one record before the measured run, and fail if the run reads fewer records than expected. Both benchmarks were repeated 5 times on the kafka-e2e Jenkins job: 10 of 10 passed and every run read all records. Logs are attached to the JIRA. Generated-by: Claude Fable 5.1 -- 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]
