lhotari commented on issue #25936: URL: https://github.com/apache/pulsar/issues/25936#issuecomment-4626383885
> One question: the residual gap where producer creation succeeds slightly past operationTimeoutMs is exactly the bug class we're fixing, just narrower. In environments where broker pod replacement DNS recovery can exceed 30s (we observed 12s, but tail latencies during AZ-spanning rollouts have been seen higher), the orphan can still form. > Would you be open to a whenComplete-style cleanup on the residual future? Diff is +5 lines, eliminates the gap entirely. > Hope thats a fair Q. > Appreciate your quick response. If the exceeds 30 seconds, the Pulsar client's operationTimeout should be increased. There shouldn't be any timeout handling in WorkerUtils.createExclusiveProducerWithRetry . In your case, since you are on an old version of Pulsar, you might be hitting multiple DNS issues that have been fixed in Pulsar / Netty. I'd strongly recommend to upgrade to a maintained Pulsar version to get the fixes. -- 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]
