helix-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "dafu (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HELIX-74) handling large number of watchers during session re-establishment
Date Wed, 27 Mar 2013 23:29:15 GMT

     [ https://issues.apache.org/jira/browse/HELIX-74?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

dafu updated HELIX-74:
----------------------

    Description: 
due to this zk bug:
https://issues.apache.org/jira/browse/ZOOKEEPER-706

we can't handling session re-establishment when watcher set size on zk client side goes beyond
1M. we need at least 2 things:
1) add watcher size monitoring mechanism, log warnings when watcher set size goes beyond 1M.
this could be done through reflection on zk-connection

2) reduce number of watchers helix needs. one possible solution is to register only one watcher
on parent node and every time we change a znode under the parent node, we touch the parent
node.

  was:
due to this zk bug:
https://issues.apache.org/jira/browse/ZOOKEEPER-706

we can't handling session re-establishment when watcher set size on zk client side goes beyond
1M. we need at 2 things:
1) add watcher size monitoring mechanism, log warnings when watcher set size goes beyond 1M.
this could be done through reflection on zk-connection

2) reduce number of watchers helix needs. one possible solution is to register only one watcher
on parent node and every time we change a znode under the parent node, we touch the parent
node.

    
> handling large number of watchers during session re-establishment
> -----------------------------------------------------------------
>
>                 Key: HELIX-74
>                 URL: https://issues.apache.org/jira/browse/HELIX-74
>             Project: Apache Helix
>          Issue Type: Bug
>            Reporter: dafu
>            Assignee: dafu
>
> due to this zk bug:
> https://issues.apache.org/jira/browse/ZOOKEEPER-706
> we can't handling session re-establishment when watcher set size on zk client side goes
beyond 1M. we need at least 2 things:
> 1) add watcher size monitoring mechanism, log warnings when watcher set size goes beyond
1M. this could be done through reflection on zk-connection
> 2) reduce number of watchers helix needs. one possible solution is to register only one
watcher on parent node and every time we change a znode under the parent node, we touch the
parent node.

--
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

Mime
View raw message