[
https://issues.apache.org/jira/browse/GEODE-9500?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17401419#comment-17401419
]
ASF subversion and git services commented on GEODE-9500:
--------------------------------------------------------
Commit 2a321cf818adae6c8399e8f407d45b17db1eb669 in geode's branch
refs/heads/develop from Darrel Schneider
[ https://gitbox.apache.org/repos/asf?p=geode.git;h=2a321cf ]
GEODE-9500: clear deltaInfo even if delta not distributed
-- deltaInfo is now cleared right after region.put is called (#6760)
-- the size tests no longer clear it
-- the unit tests for delta needed to change because the deltaInfo is now
cleared earlier
> RedisData keeps DeltaInfo references too long
> ---------------------------------------------
>
> Key: GEODE-9500
> URL: https://issues.apache.org/jira/browse/GEODE-9500
> Project: Geode
> Issue Type: Bug
> Components: redis
> Reporter: Darrel Schneider
> Assignee: Darrel Schneider
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.15.0
>
>
> The current implementation of Delta for RedisData does not clear the
> reference to the DeltaInfo until toDelta is called. This works fine in the
> normal case but in the abnormal case it can case extra memory to be used by
> the DeltaInfo instances being kept alive.
> What can happen is that if the extra server goes down so that geode is
> running with one server then some optimization are done by geode and it never
> calls toDelta. So in that case the DeltaInfo instances will be kept stored on
> the primary.
> We actually have sizing tests that found this problem but those tests were
> changes to clear the DeltaInfo reference in the test.
> Instead we should have these tests not clear it and change geode to clear it
> in AbstractRedisData.storeChanges after the put returns.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
