[jira] [Commented] (JCR-3488) JCR is taking longer time to release the lock on DMS node

["Jukka Zitting (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/JCR-3488?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13649662#comment-13649662 ] 

Jukka Zitting commented on JCR-3488:
------------------------------------

Do we have a test case that could be used to reproduce this issue?
                
> JCR is taking longer time to release the lock on DMS node
> ---------------------------------------------------------
>
>                 Key: JCR-3488
>                 URL: https://issues.apache.org/jira/browse/JCR-3488
>             Project: Jackrabbit Content Repository
>          Issue Type: Bug
>          Components: clustering, locks
>    Affects Versions: 2.2.7
>         Environment: 5 nodes Cluster environment with windows server OS and shared datastore
>            Reporter: Nilay Parmar
>
> We have implemented web based DMS using Jackrabbit(2.2.7).  This system is installed in clustered environment with windows server OS.
> Now, as we know DMS datastore is in tree structure and when we upload any document, jackrabbit locks the parent node, create a child node for the document which we want to upload and then releases the lock from parent node.
> This locking process is synchronized. If any particular parent node is locked for one document upload, other thread cannot acquire lock on that node until previous thread release the lock. Jackrabbit implementation prevents from any deadlock situation.
> As per our implementation, parent node is locked between creating a child node and set some of the properties. While putting actual content, we do not lock the parent node.
> Now, sometimes lock duration is more than 50 seconds and hence, next thread is not able to upload the document as it not getting the lock within time frame.
> This is a 5 nodes clustered environment and it requires synchronization activity, indexing activity, etc. and JCR has to do those activities in parallel but 50 seconds should be enough to complete the process and release the lock.
> We upload approximately 2000 documents < 3-5 MB every day. Plus other DMS operations are there like checkout, view, delete, etc.
> We did some testing. We tried to upload multiple documents parallel on same node with 2 different users. This creates Lock timeout issue when DMS is already loaded. Is it possible that DMS is taking longer time in other activities like Indexing, synchronization, etc. and hence locking thread not getting enough opportunity/space to release the lock?
> Please suggest the way to handle the issue.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira