[ 
https://issues.apache.org/jira/browse/KARAF-7174?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17897974#comment-17897974
 ] 

ASF GitHub Bot commented on KARAF-7174:
---------------------------------------

mattrpav commented on PR #1884:
URL: https://github.com/apache/karaf/pull/1884#issuecomment-2474150093

   > @mattrpav I see a use case: a command use a service which starts a thread. 
The service has to be hook with ssh session to stop the thread when the session 
disconnects. So the service is not a shell command but it has to clean 
resources sync with ssh session. SCR deactivate is a hook on SCR component 
lifecycle not the ssh session.
   
   What about simply implementing the JDK's Closable and follow a convention vs 
marking with a domain-specific interface? 
   
   Seems like (commandInstance instanceof Closeable) would do the trick and we 
wouldn't need to create a new interface. Thoughts?




> log:tail threads keep running after ssh exits
> ---------------------------------------------
>
>                 Key: KARAF-7174
>                 URL: https://issues.apache.org/jira/browse/KARAF-7174
>             Project: Karaf
>          Issue Type: Bug
>          Components: karaf
>    Affects Versions: 4.3.2
>         Environment: Ubuntu 20.04.2 LTS
> Clean out-of-the-box Karaf 4.3.2 distribution
>            Reporter: Gabriel Andrade
>            Assignee: Jean-Baptiste Onofré
>            Priority: Critical
>         Attachments: log-tail stack trace.txt
>
>
> When executing log:tail via Karaf's ssh connection, the created thread is not 
> killed when the ssh connection is killed or timeouts.
> I believe that all threads launched via ssh should be killed if the session 
> is no longer active.
> Steps to reproduce:
>  * Open a ssh connection with Karaf
>  * {{Run log:tail}}
>  * Either close the session without exiting log:tail or wait until it 
> timeouts in a few minutes;
>  * {{Run shell:threads | grep log:tail on the console or another ssh 
> connection}}
>  * A log tail thread should still be running.
> Every time this process is repeated, a new log:tail thread will appear and 
> stay alive.
> I've attached the return of one of my shell:thread commands.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to