activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mark Brightwell (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AMQ-3851) Infinite TemqQueue advisory messages
Date Wed, 03 Oct 2012 07:28:09 GMT

    [ https://issues.apache.org/jira/browse/AMQ-3851?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13468386#comment-13468386
] 

Mark Brightwell commented on AMQ-3851:
--------------------------------------

Hi.

I think we are seeing the same issue with fuse version 5.5.1-fuse-08-15 (was already the case
for previous versions). We are running 2 brokers is a cluster with dynamic temporary queue
forwarding. After some time, the 2 brokers get into some "vicious circle", whereby the same
temporary queues are repeatedly added then removed from each of the brokers. As Chris mentions,
this seems to use up a single thread on each process, presumably the thread removing &
creating this temporary queues. It looks like the temporary queues are simply passed back
and forth between the 2 brokers, and never get removed correctly. Only a restart of the brokers
fixes the problem.
                
> Infinite TemqQueue advisory messages
> ------------------------------------
>
>                 Key: AMQ-3851
>                 URL: https://issues.apache.org/jira/browse/AMQ-3851
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker
>    Affects Versions: 5.6.0
>            Reporter: Chris Robison
>            Assignee: Dejan Bosanac
>         Attachments: ActiveMqMessagingProvider.cs, ActiveMqRequestReceiver.cs, Log.zip,
mssd5d101-activemq.xml, mstmip103-activemq.xml, run-2012-05-30.snapshot, web-portal-code-snippet.cs
>
>
> I'm seeing a behavior that looks like a race condition of some kind. I have two brokers
(MSSD5D101, MSTMIP103). MSSD5D101 has a duplex network connector connecting it to MSTMIP103.
The two machines are in different US states so there is a VPN in between. I have two consumer
services that reply to queued messages. One consumer (WISD5P101) connects to MSSD5D101 and
the other consumer (POSD5P101) connects to MSTMIP103. I also have a web portal that sends
messages out to different queues expecting a response. The web portal, and two consumers are
all .NET based. 
> What is happening is that after some number of messages, something goes crazy and floods
the network of brokers with advisory messages pertaining to temp queues being added or removed.
By restarting one or the other of the ActiveMQ servers, things return to normal. But, while
the flood is happening, one of the machines will be at 100% CPU and become worthless. I will
be attaching logs and some code that I'm using to interact with ActiveMQ.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message