[ 
https://issues.apache.org/jira/browse/ZOOKEEPER-834?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Laxman updated ZOOKEEPER-834:
-----------------------------

    Attachment: ZOOKEEPER-834.patch

Initial patch for review

> Allow ephemeral znodes to have children created only by the owner session. 
> ---------------------------------------------------------------------------
>
>                 Key: ZOOKEEPER-834
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-834
>             Project: ZooKeeper
>          Issue Type: New Feature
>          Components: c client, java client, server
>            Reporter: Andrei Savu
>         Attachments: ZOOKEEPER-834.patch
>
>
> Ephemeral znodes are automatically removed when the client session is closed 
> or expires and this behavior makes them very useful when you want to publish 
> status information from active / connected clients. 
> But there is a catch. Right now ephemerals can't have children znodes and 
> because of that clients need to serialize status information as byte strings. 
> This serialization renders that information almost invisible to generic 
> zookeeper clients and hard / inefficient to update. 
> Most of the time the status information can be expressed as a bunch of (key, 
> value) pairs and we could easily store that using child znodes. Any ZooKeeper 
> client can read that info without the need to reverse the serialization 
> process and we can also easily update it. 
> I suggest that the server should allow the ephemeral znodes to have children 
> znodes. Each child should also be an ephemeral znode owned by the same 
> session - parent ephemeralOwner session.
> Mail Archive: 
> http://www.mail-archive.com/zookeeper-dev@hadoop.apache.org/msg09819.html
> Another discussion about the same topic:
> http://www.mail-archive.com/zookeeper-dev@hadoop.apache.org/msg08165.html

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

        

Reply via email to