[
https://issues.apache.org/jira/browse/RATIS-1921?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Andika updated RATIS-1921:
-------------------------------
Description:
When benchmarking Ozone streaming pipeline using `ozone freon ockg`, the
benchmark would not end although all the keys have been written.
Using Arthas, found that non-daemon threads like
"NettyClientStreamRpc-workerGroup–thread1" are still running even after the
benchmark has finished.
It seems that when hdds.ratis.raft.netty.dataStream.client.worker-group.share
is set to true, it will create a worker group that will be shared. However,
this worker group will never be closed causing the JVM shutdown hook to never
be triggered. The benchmark was able to work normally if the share
configuration is disabled.
Need to find a way to shutdown the worker group gracefully without losing the
the shared worker group benefit.
was:
When benchmarking Ozone streaming pipeline using `ozone freon ockg`, the
benchmark would not end although all the keys have been written.
Using Arthas, found that non-daemon threads like
"NettyClientStreamRpc-workerGroup–thread1" are still running even after the
benchmark has finished.
It seems that when hdds.ratis.raft.netty.dataStream.client.worker-group.share
is set to true, it will create a worker group that will be shared. However,
this worker group will never be closed causing the JVM shutdown hook to never
be triggered. The benchmark was able to work normally if the share
configuration is disabled.
Need to find a way to shutdown the worker group gracefully.
> Close shared worker group in WorkerGroupGetter
> ----------------------------------------------
>
> Key: RATIS-1921
> URL: https://issues.apache.org/jira/browse/RATIS-1921
> Project: Ratis
> Issue Type: Sub-task
> Reporter: Ivan Andika
> Assignee: Ivan Andika
> Priority: Minor
>
> When benchmarking Ozone streaming pipeline using `ozone freon ockg`, the
> benchmark would not end although all the keys have been written.
> Using Arthas, found that non-daemon threads like
> "NettyClientStreamRpc-workerGroup–thread1" are still running even after the
> benchmark has finished.
> It seems that when hdds.ratis.raft.netty.dataStream.client.worker-group.share
> is set to true, it will create a worker group that will be shared. However,
> this worker group will never be closed causing the JVM shutdown hook to never
> be triggered. The benchmark was able to work normally if the share
> configuration is disabled.
> Need to find a way to shutdown the worker group gracefully without losing the
> the shared worker group benefit.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
