[
https://issues.apache.org/jira/browse/SLING-12690?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17936389#comment-17936389
]
Timothee Maret commented on SLING-12690:
----------------------------------------
Potentially by sampling stack traces and deciding that the import is stuck when
the commit stack trace has not changed for a configured number of samples and
the thread is runnable. That would be complex though. So I think that a 3h
timeout, without sampling the stacks, is probably a better first step.
> Skip package if import is stuck for too long
> --------------------------------------------
>
> Key: SLING-12690
> URL: https://issues.apache.org/jira/browse/SLING-12690
> 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
>
>
> When importing a content package we call filevault to import the package into
> oak.
> This is a synchronous call that blocks until the import is finished.
> We have cases where this import takes much longer than expected and causes
> unavailability of replication for other authors.
> We should introduce a maximum time after which we consider the import to be
> failed and mark the package as skipped.
> So if an import takes longer than this defined time we must:
> * Send out a status message to skip the package. So other pods also skip the
> package
> * Mark the offset of the package as processed
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
