Closed vaalkor closed 1 year ago
Hi! Thank you for reporting the issue. I'll look into the issue within several days. In 2.9 there was a major change in task library, this might be the cause. Right now task builders are context-sensitive, and we'll be able to move to context insensitive versions soon, since the fix landed 4 days ago. So once new fsharp compiler version is out (.net 7 rc2) , I'll release the version of Pulsar.Client with context-insensitive tasks anyway, and we'll be able to test your case again - will it help or not.
Interesting. That sounds like a plan. Thanks!
@vaalkor I think I've found the issue, published 2.10.3 with the fix, please test if it works for you
Closing as abandoned
Hi, sorry I didn't reply to this. Getting pulled in many directions at work and forgot about it! I did run it again and the deadlock seemed to be gone but things seemed quite slow. I didn't look too deeply into it though because of other concerns. I will have to do some proper testing and get back to you.
@vaalkor I see, you are welcome to create another issue about the slowness concern.
Hi there! I seem to be encountering a consumer hang when running this test project: https://github.com/vaalkor/PulsarTest
It hangs after printing this a few times: warn: PulsarLogger[0] mt/consumer(0, b016a) UnackedTracker 230 messages have timed-out
It hangs with pulsar.client 2.9 onwards. If I downgrade to 2.8.1 there's no hang. Turning off EnableRetry on the consumer changes the behaviour a lot. I'm not really sure what's happening here but I'd appreciate if you could shed any light on the matter. Thanks!
I'm running pulsar in docker and I've tried it with 2.8.3, 2.9.3, 2.10.1 and they all seem to behave similarly.