[jira] [Commented] (HBASE-2459) Use finer-grain locking in HConnectionManager

["Nicolas Spiegelberg (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-2459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13009429#comment-13009429 ] 

Nicolas Spiegelberg commented on HBASE-2459:
--------------------------------------------

@Stack:  I think this is different from my change in HBASE-3653.  This jira is related to the region cache, not the RegionServer cache.  I believe the nasty culprit in this jira is currently named:
{code}
  synchronized (regionLockObject) 
{code}
Talked with jgray and I think we can downgrade this issue from Critical to Minor.

> Use finer-grain locking in HConnectionManager
> ---------------------------------------------
>
>                 Key: HBASE-2459
>                 URL: https://issues.apache.org/jira/browse/HBASE-2459
>             Project: HBase
>          Issue Type: Improvement
>          Components: client
>            Reporter: Jonathan Gray
>            Assignee: Nicolas Spiegelberg
>            Priority: Critical
>              Labels: moved_from_0_20_5, noob
>             Fix For: 0.92.0
>
>
> Currently we synchronize within HConnectionManager.locateRegionInMeta() when looking up the cached location (and querying meta if not cached).  We use the same lock for every user-space region.
> We really only need per-region synchronization here.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira