[
https://issues.apache.org/jira/browse/HBASE-9917?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13817396#comment-13817396
]
stack commented on HBASE-9917:
------------------------------
Any chance of a review here?
Previous we had hard-pegged 256 threads just sitting there idle. Now we have
only threads that are doing work or those waiting around a while (keepalive 10
seonds) to see if more work about to come (else they die).
> Fix it so Default Connection Pool does not spin up max threads even when not
> needed
> -----------------------------------------------------------------------------------
>
> Key: HBASE-9917
> URL: https://issues.apache.org/jira/browse/HBASE-9917
> Project: HBase
> Issue Type: Sub-task
> Components: Client
> Reporter: stack
> Assignee: stack
> Fix For: 0.98.0, 0.96.1
>
> Attachments: 9917.txt, pool.txt
>
>
> Testing, I noticed that if we use the HConnection executor service as opposed
> to the executor service that is created when you create an HTable without
> passing in a connection: i.e
> HConnectionManager.createConnection(config).getTable(tableName)
> vs
> HTable(config, tableName)
> ... then we will spin up the max 256 threads and they will just hang out
> though not being used.
> We are encouraging HConnection#getTable over new HTable so worth fixing.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
