Lari Hotari

> https://bugs.openjdk.org/browse/JDK-8330017 > > @lhotari @BewareMyPower We've run into this issue several times. The ForkJoinPool may stop executing tasks after running for a while, and we eventually confirmed that the...

> @lhotari > Java 21.0.8 fixed the bug, it seems Java 21 also affected. > https://bugs.openjdk.org/browse/JDK-8351933 @zhanghaou Thanks for sharing. It looks like it's also fixed in 17.0.17 `apachepulsar/pulsar:4.0.6` contains...

> For users deploying with Docker, this might be helpful. However, many users are likely installing the JDK themselves and running Pulsar directly on a VM. As for the JDK...

> Thank you for providing the additional information. I think it would be better to use a dedicated ForkJoinPool where necessary, instead of relying on the default common pool everywhere,...

Java 17.0.17 has been released. This change is no longer needed as a workaround for the Java bug. I'll initiate 3.0.15 release so that we'll have Java 17.0.17 used in...

There's #23980 in this area, however I don't know if it will prevent the issue.

> I think [#23980](https://github.com/apache/pulsar/pull/23980) will not prevent the issue because it neither prevents a compaction and topic deletion from running simultaneously nor interrupts a compaction when the topic is deleted...

potentially related PR #24366

> Currently, there are two clusters that have set up geo namespace replica synchronization, but in one of the clusters, it is found that the messages synchronized from the cluster...

> Or it is not that it is not out of synchronization, but that the delay is relatively large. Is this normal? The following is the storage monitoring of two...