zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Erwin Tam (JIRA)" <j...@apache.org>
Subject [jira] Commented: (ZOOKEEPER-886) Hedwig Server stays in "disconnected" state when connection to ZK dies but gets reconnected
Date Tue, 05 Oct 2010 20:58:33 GMT

    [ https://issues.apache.org/jira/browse/ZOOKEEPER-886?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12918178#action_12918178

Erwin Tam commented on ZOOKEEPER-886:

Bug was an error in the logic of the if condition.  For the ZK reconnect condition, the if
logic would bypass the part to change the server back to the "not suspended" state.  Modified
this so it will do it correctly now.

> Hedwig Server stays in "disconnected" state when connection to ZK dies but gets reconnected
> -------------------------------------------------------------------------------------------
>                 Key: ZOOKEEPER-886
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-886
>             Project: Zookeeper
>          Issue Type: Bug
>          Components: contrib-hedwig
>            Reporter: Erwin Tam
>            Assignee: Erwin Tam
>         Attachments: ZOOKEEPER-886.patch
> The Hedwig Server is connected to ZooKeeper.  In the ZkTopicManager, it registers a watcher
so that if it ever gets disconnected from ZK, it will temporarily fail all incoming requests
since the Hedwig server does not know for sure if it is still the master for the topics. 
When the ZK client gets reconnected, the logic currently is wrong and it does not unset the
suspended flag.  Thus once it gets disconnected, it will stay in the suspended state forever,
thereby making the Hedwig server hub dead.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message