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 "Marcel Reutegger (JIRA)" <ji...@apache.org> on 2015/04/15 17:14:58 UTC

[jira] [Commented] (OAK-2768) Consider fair mode for backgroundOperationLock

    [ https://issues.apache.org/jira/browse/OAK-2768?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14496323#comment-14496323 ] 

Marcel Reutegger commented on OAK-2768:
---------------------------------------

Ran ObservationTest with 16 writers and a save interval of 10 nodes with fair and non-fair mode of the backgroundOperationLock. The attached 'lock time.png' shows the time it took to acquire the lock for each of the test runs. At least for this test, there is not much difference between the two modes. 

> Consider fair mode for backgroundOperationLock
> ----------------------------------------------
>
>                 Key: OAK-2768
>                 URL: https://issues.apache.org/jira/browse/OAK-2768
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: core, mongomk
>            Reporter: Marcel Reutegger
>            Assignee: Marcel Reutegger
>             Fix For: 1.3.0
>
>         Attachments: lock time.png
>
>
> The backgroundOperationLock in DocumentNodeStore uses the default non-fair acquisition order. According to JavaDoc of ReentrantReadWriteLock it is possible that a background operation task gets delayed for a long time when the system is under load. We should probably consider using the fair mode for the backgroundOperationLock to make sure background operation tasks do not get delayed excessively.



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