tolbertam commented on PR #2025: URL: https://github.com/apache/cassandra-java-driver/pull/2025#issuecomment-3275103789
thanks for the heads up @jasonk000 (FYI @absurdfarce) > The scenario is the logger throwing unanticipated exceptions (when we are nearing OOM) at a time which messes up the counter state. We can fix this by tracking and unwinding at the right moment and ensuring correct operation ordering before calling the logger. Oh that sounds fun, OOMs causing runtime exceptions to be thrown in unexpected places can definitely screw things up in unpredictable ways. I suppose since this PR added some logging that might explain why hadn't seen this before? Although I do see there was logging previously in some places. > Secondly, we found an issue that hampered the investigation -- since we are pushing more load now, the recursion in the request-done-and-submit-next flow is leading to very deep stacks, so we will be switching that to a loop to keep the stack manageable. Is that a new issue or something that would have previously existed prior to this PR? Has been a while since I looked at this PR so would need to context set, but don't recall this changing how things are called, so guessing that would have been a previous problem? Or is it something subtle in the queuing changes that contribute? -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
