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 2014/02/13 21:21:24 UTC

[jira] [Commented] (OAK-1422) Large number of queued EventGenerator.Continuation on big transactions

    [ https://issues.apache.org/jira/browse/OAK-1422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13900696#comment-13900696 ] 

Michael Dürig commented on OAK-1422:
------------------------------------

A first test showed that committing 5M node in a single transaction succeeds without significant memory overhead when processing the content diff in a depth first manner, while when processing it in a breath first manner Oak runs out of heap space. 

> Large number of queued EventGenerator.Continuation on big transactions
> ----------------------------------------------------------------------
>
>                 Key: OAK-1422
>                 URL: https://issues.apache.org/jira/browse/OAK-1422
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>          Components: core
>            Reporter: Michael Dürig
>
> The queue of {{EventGenerator.Continuation}} grows in the order of changed number of leave nodes instead of in the order of the depth of the changed sub tree. 
> This is caused by the {{EventGenerator}} processing events in a breath first manner causing events from each level of the tree to be held in the tree until all lower levels have been processed. I suggest to change the processing order to depth first, which should significantly lower the upper bound of the number of queued {{Continuation}} instances. 



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)