[
https://issues.apache.org/jira/browse/NIFI-7286?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tamas Palfy reassigned NIFI-7286:
---------------------------------
Assignee: Tamas Palfy
> ListenTCPRecord doesn't release port when stopping + terminating while running
> ------------------------------------------------------------------------------
>
> Key: NIFI-7286
> URL: https://issues.apache.org/jira/browse/NIFI-7286
> Project: Apache NiFi
> Issue Type: Bug
> Reporter: Tamas Palfy
> Assignee: Tamas Palfy
> Priority: Major
>
> *Reproduction steps*
> # Set '{{ListenTCPRecord.Read Timeout}}' property to manageable value (like
> 120 second)
> # Start {{ListenTCPRecord}}, listen on port XXX
> # {{telnet HOST XXX}} (This will open and keep a connection to that port)
> # Stop {{ListenTCPRecord}} while connection is open (After that, notice the 2
> active threads on the processor)
> # Terminate {{ListenTCPRecord}}
> # Now any process (be it {{ListenTCPRecord}} or {{ListenTCP}}) that tries to
> listen on port XXX will report an {{"Address already in use"}} error
> *Root cause*
> {{ListenTCPRecord}} is trying to do cleanup in an {{@OnStopped}} method. The
> problem is that these methods won't be called while there are still running
> worker threads. When the processor is terminated, it eliminates the cleanup
> thread as well, so the cleanup never happens.
> {{ListenTCP}} for example doesn't exhibit this issue because its cleanup
> method is not annotated with {{@OnStopped}} but with {{@OnUnscheduled}}.
> *Fix*
> The solution is fairly simple: we just need to annotate the
> {{ListenTCPRecord}} cleanup method with {{@OnUnscheduled}} instead of
> {{@OnStopped}}.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
