[jboss-jira] [JBoss JIRA] Commented: (JBCACHE-772) State transfer not handled properly in case of reconnect

Brian Stansberry (JIRA) jira-events at lists.jboss.org
Fri Apr 25 10:14:09 EDT 2008


    [ http://jira.jboss.com/jira/browse/JBCACHE-772?page=comments#action_12410735 ] 
            
Brian Stansberry commented on JBCACHE-772:
------------------------------------------

It's not a situation due to a merge. It's a case where the cache's channel gets suspected and reconnects.

Conceptually very similar to a merge though in that the node's state has become inconsistent with respect to the cluster, so I could see how a fix would use the same logic as the JBCACHE-471 solution.  I suppose what I wrote in the description is  a policy of "throw away my state and accept the group's state" which is not always the right answer.

> State transfer not handled properly in case of reconnect
> --------------------------------------------------------
>
>                 Key: JBCACHE-772
>                 URL: http://jira.jboss.com/jira/browse/JBCACHE-772
>             Project: JBoss Cache
>          Issue Type: Bug
>      Security Level: Public(Everyone can see) 
>          Components: Clustering
>    Affects Versions: 1.4.0.SP1, 1.4.0.GA, 1.3.0.SP3, 1.3.0.SP2, 1.3.0.SP1, 1.3.0.GA, 1.2.4SP2, 1.2.4SP1, 1.2.4
>            Reporter: Brian Stansberry
>         Assigned To: Manik Surtani
>             Fix For: 2.2.0.GA
>
>
> In createService(), TreeCache set Channel.AUTO_GET_STATE to true.  This should only be done if the cache is configured to do an initial state transfer.
> That's easy to fix.  What's harder to fix is what to do when region-based marshalling is in effect. Probably, something like the following:
> 1) In viewAccepted, monitor for the case where a view is issue that doesn't include the current node.  Set a flag and ??? (throw exceptions when attempts are made to use the cache???)
> 2) When another view is accepted that once again includes the current node, iterate through the various marshalling regions, inactivating and then reactivating them (reactivation will cause a partial state transfer).

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        



More information about the jboss-jira mailing list