activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Timothy Bish (JIRA)" <j...@apache.org>
Subject [jira] [Resolved] (AMQ-3450) Acknowledging prefetched messages using client acknowledge mode
Date Sat, 20 Aug 2011 21:26:27 GMT

     [ https://issues.apache.org/jira/browse/AMQ-3450?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Timothy Bish resolved AMQ-3450.
-------------------------------

    Resolution: Not A Problem

Fixed by adding a wait for receipt on the last message sent so that all messages get processed
before the socket is broken.  

> Acknowledging prefetched messages using client acknowledge mode
> ---------------------------------------------------------------
>
>                 Key: AMQ-3450
>                 URL: https://issues.apache.org/jira/browse/AMQ-3450
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker
>            Reporter: Hariharan
>              Labels: broker
>         Attachments: my_client.pl
>
>
> We have a situation in front of us and its explained as below:
> 1. We have durable subscriber subscribed to a topic. 
> 2. This durable subscriber is a perl script which is run by a daemon.
> 3. The perl script uses stomp to connect to the broker.
> 4. The perl script wakes up every 5 mins, checks for messages in the topic and processes
them in a batch by pre-fetching the messages.
> 5. The subscriber uses a client acknowledgement and acknowledges only the last message
of the batch.
> 6. We are using AMQ 5.5 with kahaDB persistence.
> Now what we see is,
> 1. Even though the messages are processed in a batch and the last message is acknowledged
the inflight count does not come down.
> 2. Enqueue count, Dequeue count and Dispatch count do not match.
> 3. The journal files are not getting cleaned up.
> I do understand that the journal files would be cleaned up once the references to the
messages are lost or removed (i.e. the messages are consumed). But does it have to do anything
with the various count attributes I see on the topic?
> Also should I expect the inflight count to come down to 0 if client crashes and then
consumes all messages after come back live?
> I also see that dequeue count stays at 0 even when the dispatch count and the inflight
counts change. I believe dequeue count has got a direct relation with messages getting removed
from the topic.
> Please let me know if there could be any other reason that could cause the journal files
to stay back.
> Thank you
> Hari

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message