ableegoldman commented on pull request #10409: URL: https://github.com/apache/kafka/pull/10409#issuecomment-817035420
Thanks @showuon . I guess my question is really this: why is it that this test, and no others, is frequently flaky due to dropping out on the session interval? There's nothing really "special" about it, ie it uses a single StreamThread and the default hb/session interval. It's hard to believe that Jenkins is so bad that a single consumer can't get a single heartbeat to the brokers in 10s. Maybe there is a bug or other slowdown in the networking layer, I don't know -- my point was really that dropping out on the heartbeat should not be expected, and if that's really causing tests to fail then it seems like a valid problem to investigate further, and not necessarily just expected flakiness (Even if it's not what this particular test was _supposed_ to be testing) Does that make sense? I do see that it appears to drop out on hb expiration, but that only occurs twice so I wonder if that's really causing the flakiness or whether it should be able to recover from this. What if we just merge the other improvements in this PR to start with, and see how much that helps? -- 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: us...@infra.apache.org
