You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-issues@jackrabbit.apache.org by "Michael Dürig (JIRA)" <ji...@apache.org> on 2016/09/13 08:26:21 UTC

[jira] [Commented] (OAK-4731) Refine forced compaction

    [ https://issues.apache.org/jira/browse/OAK-4731?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15486644#comment-15486644 ] 

Michael Dürig commented on OAK-4731:
------------------------------------

The option to force compaction in {{SegmentNodeStoreService}} has been replaced by an option to set a time out on forced compaction ({{compaction.force.timeout}}). That time out specifies the number of seconds to attempt forced compaction. This includes the time it takes to acquire a exclusive write lock on the node store and the compaction time itself. Forced compaction will not be attempted if the time out value is set to 0.

> Refine forced compaction
> ------------------------
>
>                 Key: OAK-4731
>                 URL: https://issues.apache.org/jira/browse/OAK-4731
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: segment-tar
>            Reporter: Michael Dürig
>            Assignee: Michael Dürig
>              Labels: compaction, gc
>             Fix For: Segment Tar 0.0.12
>
>
> Forced compaction currently acquires an exclusive write lock on the repository blocking all concurrent commits during the complete time it needs to finish compaction. I think we should refine this:
> * Add a time out so we could limit the time during which the repository does not accept writes while still giving compaction another chance to finish.
> * Boost the compaction threads priority. This could actually already be done during the regular compaction cycles to increase the changes to finish in time. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)