[
https://issues.apache.org/jira/browse/GEODE-2367?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15842344#comment-15842344
]
ASF subversion and git services commented on GEODE-2367:
--------------------------------------------------------
Commit 072413fe2cb86ddf5093548d7c847b43c0775811 in geode's branch
refs/heads/feature/GEODE-2367 from [~huynhja]
[ https://git-wip-us.apache.org/repos/asf?p=geode.git;h=072413f ]
GEODE-2367: Lucene indexes do not handle ha scenarios
* Added afterSecondary callback to partition listener to allow cleaning up
of the index repo when the bucket losses primary
* Added lock prior to creating the bucket indexes to prevent multiple index
writers from being available at a time
* Changed single point of lucene index creation, no longer creating on the fly
> Lucene Index does not handle ha scenarios
> -----------------------------------------
>
> Key: GEODE-2367
> URL: https://issues.apache.org/jira/browse/GEODE-2367
> Project: Geode
> Issue Type: Bug
> Components: lucene
> Reporter: Jason Huynh
> Assignee: Jason Huynh
>
> In ha scenarios or during rebalance, the lucene index listeners allow
> multiple nodes to write to the file/chunk regions.
> The Index repositories are also not being cleaned up when a bucket goes from
> primary to secondary.
> This work will need to refactor the creation of the Lucene index to not
> create one on the "fly" when a get/or query function is executed. Otherwise
> a function execution on a node that has just been turned secondary, may cause
> it to create an index repo that it should not create.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
