[
https://issues.apache.org/jira/browse/HBASE-6732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13467218#comment-13467218
]
stack commented on HBASE-6732:
------------------------------
Looking at the patch, it is a long way from trunk. It is doing compression on
the send -- which we don't do yet -- and its doing it up in byte array first
before we enter the synchronize block. Trunk on other hand is all pb'd. We
are doing the build inside the synchronize which seems unnecessary. We should
change this fact at least. I'm also wondering if these size computes are not
redoing build work. Will look at pb optimizations over in hbase-6910
> [89-fb] Reduce scope of synchronized block in
> HBaseClient.Connection#sendParam.
> -------------------------------------------------------------------------------
>
> Key: HBASE-6732
> URL: https://issues.apache.org/jira/browse/HBASE-6732
> Project: HBase
> Issue Type: Test
> Components: Client
> Reporter: Michal Gregorczyk
> Assignee: Michal Gregorczyk
> Fix For: 0.96.0
>
> Attachments:
> 0002-jira-HBASE-6732-89-fb-Reduce-scope-of-synchronized-b.patch
>
>
> sendParam method synchronizes on out stream (so that only one thread writes
> to socket). Right now it prepares (eg. compress) message under this lock as
> well. Lock should be taken only for sending.
--
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
