[
https://issues.apache.org/jira/browse/SLING-10112?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17276956#comment-17276956
]
Timothee Maret commented on SLING-10112:
----------------------------------------
Sending commands and importing packages are required to be two independent and
parallel processes.
The approach in the PR #65 postpones the package consumer initialisation to
when the command consumer is idle. The idle check is an heuristic. When the
heuristic fails, new attempts can be triggered by restarting the process. Using
an heuristic here seems appropriate considering the alternative would require
persisting yet another state in the repository. Looking at the PR, I think we
should make sure the idle check times out eventually.
> Distribution Queue can not be reliably cleared due to race condition
> processing delete messages
> -----------------------------------------------------------------------------------------------
>
> Key: SLING-10112
> URL: https://issues.apache.org/jira/browse/SLING-10112
> Project: Sling
> Issue Type: Improvement
> Components: Content Distribution
> Reporter: Christian Schneider
> Assignee: Christian Schneider
> Priority: Major
>
> Replication queue can not be reliable cleared due to race condition with
> processing the 'delete' messages.
> If the first message is stuck during processing, the 'clear queue' signal
> might never have effect on the subscriber.
> A restart of the process does not help as the same situation might occur.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
