[GitHub] [kafka] ableegoldman edited a comment on pull request #8963: KAFKA-10017: fix flaky EosBetaUpgradeIntegrationTest

[GitBox <gi...@apache.org>]

ableegoldman edited a comment on pull request #8963:
URL: https://github.com/apache/kafka/pull/8963#issuecomment-652773774


   Looks like it only failed one time across all three builds in the last run. But I pushed some fixes for the remaining flakiness:
   1. Use the IntegrationTestUtils `#getStore` to retry on InvalidStateStoreException
   2. Relax the assignment listener check: the way we currently make sure to reach a stable assignment is to save the current number of stable assignments before a rebalance-triggering event (eg 2nd client is stopped) and then wait for the number of stable assignments to increase by exactly one. In some cases the test failed because we actually got to more than the expected number of stable assignments. I'm not exactly sure why, but I don't think it's necessarily a problem and we can probably just relax the condition to make sure we had at least one more stable assignment
   
   Can you kick off the tests again @guozhangwang ? I'm interested to see if the fix for 2. is enough or if whatever is causing the extra rebalance will surface as a test failure in another way


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org