[
https://issues.apache.org/jira/browse/ARTEMIS-2188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16745597#comment-16745597
]
ASF subversion and git services commented on ARTEMIS-2188:
----------------------------------------------------------
Commit 9c1ecd2558e09cbe439814dbf96641f06916a315 in activemq-artemis's branch
refs/heads/2.6.x from yang wei
[ https://gitbox.apache.org/repos/asf?p=activemq-artemis.git;h=9c1ecd2 ]
ARTEMIS-2188 fix address size leak caused by large page message
(cherry picked from commit 52d236d850aeca10a49aa54833141df2f51722a2)
> Paged large message results in pagingStore sizeInBytes leak when the page is
> evicted
> ------------------------------------------------------------------------------------
>
> Key: ARTEMIS-2188
> URL: https://issues.apache.org/jira/browse/ARTEMIS-2188
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Components: Broker
> Reporter: yangwei
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> When we write a paged large message,
> addLiveMessage->incrementDelayDeletionCount->
> incrementRefCount->onDurableUp is called where pagingstore size and global
> size is added.
> The page where the large message resides maybe evicted from cache.
> When the page is deleted, LargeServerMessageImpl::decrementDelayDeletionCount
> is called the last time but the large message is not the original one which
> means context is null and nonDurableDown is not called resulting in
> pagingstore size and global size not properly subtracted.
> Writing large paged message should be handled like regular paged message. The
> large paged message doesn't account for the memory size, we don't need to
> call PagingStoreImpl::onDurableUp.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
