You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zookeeper.apache.org by "Hadoop QA (JIRA)" <ji...@apache.org> on 2011/03/26 00:03:05 UTC

[jira] [Commented] (ZOOKEEPER-1030) Increase default for maxClientCnxns

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-1030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011474#comment-13011474 ] 

Hadoop QA commented on ZOOKEEPER-1030:
--------------------------------------

-1 overall.  Here are the results of testing the latest attachment 
  http://issues.apache.org/jira/secure/attachment/12474672/zk-1030.txt
  against trunk revision 1082362.

    +1 @author.  The patch does not contain any @author tags.

    -1 tests included.  The patch doesn't appear to include any new or modified tests.
                        Please justify why no new tests are needed for this patch.
                        Also please list what manual steps were performed to verify this patch.

    -1 patch.  The patch command could not apply the patch.

Console output: https://hudson.apache.org/hudson/job/PreCommit-ZOOKEEPER-Build/202//console

This message is automatically generated.

> Increase default for maxClientCnxns
> -----------------------------------
>
>                 Key: ZOOKEEPER-1030
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1030
>             Project: ZooKeeper
>          Issue Type: Improvement
>    Affects Versions: 3.2.2
>            Reporter: Todd Lipcon
>            Priority: Trivial
>             Fix For: 3.2.3
>
>         Attachments: zk-1030.txt
>
>
> The default for maxClientCnxns is 10, which is too low for many applications. For example, HBase users often run MR jobs where each task needs to use ZooKeeper to talk to HBase. This means that each slot on the tasktracker will have at least one ZK connection. With today's beefy machines, that's easily 20+ connections per node.
> I would suggest bumping the default to 60, which will still protect against runaway nodes (eg a leak in a tight loop) but won't impact MR jobs that need to talk to ZK.

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