[
https://issues.apache.org/jira/browse/HBASE-14433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14744492#comment-14744492
]
Enis Soztutar commented on HBASE-14433:
---------------------------------------
Is this meta lookup threads or regular core threads or both that we want to
tune down?
We have this for meta lookup:
{code}
conf.getInt("hbase.hconnection.meta.lookup.threads.max", 128),
conf.getInt("hbase.hconnection.meta.lookup.threads.core", 10),
{code}
and this for the regular pool:
{code}
this.batchPool =
getThreadPool(conf.getInt("hbase.hconnection.threads.max", 256),
conf.getInt("hbase.hconnection.threads.core", 256), "-shared-",
null);
{code}
Agreed that 256 core seems a bit stretched.
> Set down the client executor core thread count from 256 to number of
> processors
> -------------------------------------------------------------------------------
>
> Key: HBASE-14433
> URL: https://issues.apache.org/jira/browse/HBASE-14433
> Project: HBase
> Issue Type: Sub-task
> Components: test
> Reporter: stack
> Assignee: stack
> Attachments: 14433.txt
>
>
> HBASE-10449 upped our core count from 0 to 256 (max is 256). Looking in a
> recent test run core dump, I see up to 256 threads per client and all are
> idle. At a minimum it makes it hard reading test thread dumps. Trying to
> learn more about why we went a core of 256 over in HBASE-10449. Meantime will
> try setting down configs for test.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
