[infinispan-issues] [JBoss JIRA] (ISPN-2475) When L1.onRehash is enabled, L1 invalidations should be sent to the previous owners
RH Bugzilla Integration (JIRA)
jira-events at lists.jboss.org
Thu Oct 31 06:37:02 EDT 2013
[ https://issues.jboss.org/browse/ISPN-2475?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12826091#comment-12826091 ]
RH Bugzilla Integration commented on ISPN-2475:
-----------------------------------------------
Radim Vansa <rvansa at redhat.com> made a comment on [bug 1025258|https://bugzilla.redhat.com/show_bug.cgi?id=1025258]
When L1.onRehash is enabled (by default) and the node loses ownership of some keys, it does not remove them immediately but keeps them in L1. New owners should record this node in L1ManagerImpl.requestors in order to send L1 invalidation when the key is overwritten. However, this is not implemented.
The result is that node may read outdated entry (the cluster is not consistent).
There's a comment about the intended behaviour in StateConsumerImpl, line 385.
> When L1.onRehash is enabled, L1 invalidations should be sent to the previous owners
> -----------------------------------------------------------------------------------
>
> Key: ISPN-2475
> URL: https://issues.jboss.org/browse/ISPN-2475
> Project: Infinispan
> Issue Type: Bug
> Components: State transfer
> Affects Versions: 5.1.0.FINAL, 6.0.0.CR1
> Reporter: Dan Berindei
> Assignee: William Burns
> Fix For: 6.1.0.Final
>
>
> Copied from the parent:
> {quote}
> [...] we can keep track of the consistent hashes in the last 10 minutes (or whatever the L1 lifespan is) and the time of the last invalidation sent for each key. When we need to send a new invalidation, we add the owners in all the consistent hashes since the last invalidation to the invalidation command recipients.
> {quote}
--
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
More information about the infinispan-issues
mailing list