[ 
https://issues.apache.org/jira/browse/SLING-12691?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Christian Schneider updated SLING-12691:
----------------------------------------
    Affects Version/s:     (was: Content Distribution Journal Core 0.5.2)

> Make sure clear commands are executed even when there is an import running
> --------------------------------------------------------------------------
>
>                 Key: SLING-12691
>                 URL: https://issues.apache.org/jira/browse/SLING-12691
>             Project: Sling
>          Issue Type: Improvement
>          Components: Content Distribution
>            Reporter: Christian Schneider
>            Assignee: Christian Schneider
>            Priority: Major
>             Fix For: Content Distribution Journal Core 0.5.2
>
>
> We had issues where a user tried to clear a queue but it did not have any 
> effect.
> What happens is that the clear action triggers a clear command on command 
> topic.
> The problem is that publish only skips packages based on the clear command if 
> no current import is running.
> So if an import runs for very long or blocks then the clear command is not 
> executed. If then the pod is restarted there is a race condition where it can 
> happen that the same import is started before the pod receives the clear 
> command. So it gets stuck again in the same way and can not work on the clear 
> command.
> The result is that the queue shows as non moving and the user has the 
> experience that the clear command had no effect.
> ---
> This issue aims to update the offset according to the clear command and send 
> a status message of type skip even if the import is still running. 
> If then the pod is restarted it will have skipped the package already and 
> work fine.
> There is one downside to this:
> If the import is just long running but succeeds in the end then the pod 
> importing the package will have the package but other pods might have skipped 
> it.
> This case will be fixed after next compaction run though so I think it is 
> acceptable.



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

Reply via email to