[
https://issues.apache.org/jira/browse/AMQ-5960?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15956982#comment-15956982
]
Justin Reock commented on AMQ-5960:
-----------------------------------
I don't believe this issue is fixed, it is still reproducible in 5.14.4. Same
test case, create durable subs with persistent messages, connect consumers and
start draining. Hard-kill the broker and restart it, allow consumers to drain.
There will be "lingering" pendingQueueSize counts, although there is no actual
message loss. KahaDB just doesn't update the ack position. Tested with
LevelDB and cannot recreate, this only happens with Kaha.
> Durable subscriber pendingQueueSize wrong after force shutdown of the broker
> ----------------------------------------------------------------------------
>
> Key: AMQ-5960
> URL: https://issues.apache.org/jira/browse/AMQ-5960
> Project: ActiveMQ
> Issue Type: Bug
> Affects Versions: 5.12.0
> Reporter: Dejan Bosanac
> Assignee: Dejan Bosanac
> Fix For: 5.13.0
>
>
> The main reason for this is that ackPositions (based on which
> pendingQueueSize is calculated) is not rebuilt correctly during index recover.
> To reproduce it:
> 1. Start the broker
> 2. Start producer and durable consumer on a topic, making a load
> 3. Kill the broker with -9
> 4. Restart the broker
> The durable consumer pendingQueueSize is off by one and is never corrected
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
