jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Joerg Hoh (Jira)" <j...@apache.org>
Subject [jira] [Commented] (OAK-9321) Oak sessions should be refreshed before tail compaction
Date Thu, 12 Aug 2021 18:57:00 GMT

    [ https://issues.apache.org/jira/browse/OAK-9321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17398238#comment-17398238

Joerg Hoh commented on OAK-9321:

[~wunsch] It is not implemented. You refer to a feature, which logs unclosed Sling ResourceResolvers,
but that's it. It does not refresh those ResourceResolvers (or their enclosed JCR sessions).
Also these ResourceResolvers are only logged when their respective finalizers are called,
and that's unrelated to the process if compaction.

> Oak sessions should be refreshed before tail compaction
> -------------------------------------------------------
>                 Key: OAK-9321
>                 URL: https://issues.apache.org/jira/browse/OAK-9321
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: jcr
>    Affects Versions: 1.22.5
>            Reporter: Joerg Hoh
>            Priority: Major
> Although discouraged since forever, long running JCR sessions are still seen quite often
in applications I see.
> One effect of these long running sessions is that in combination with tail compaction
it can lead to SegmentNotFoundExceptions. These exceptions are not persistent, meaning that
the storage layer is not affected. A restart will cause the SNFE to vanish until they appear
again after some time.
> Of course fixing the long running sessions is definitely the right way to fix this problem,
but it'S unlikely that it can be fixed easily in all cases. Therefor I would propose to have
a workaround, which refreshes all open sessions (older than 1h?) before performing tail compaction.
It would be great, if the sessions would be logged, which are refreshed, so it can serve as
input for the fixing.

This message was sent by Atlassian Jira

View raw message