activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Timothy Bish (Commented) (JIRA)" <jira+amq...@apache.org>
Subject [jira] [Commented] (AMQNET-371) NMSConnectionException does not trigger failover recovery.
Date Wed, 29 Feb 2012 00:56:12 GMT

    [ https://issues.apache.org/jira/browse/AMQNET-371?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13218783#comment-13218783
] 

Timothy Bish commented on AMQNET-371:
-------------------------------------

First thing to do is figure out what the exception is.  That would help to determine if we
aren't doing something right in failover.  The fix you put it looks like it could cause some
oddness in sessions where clients are in Client ack or in transactions as it dumps the consumers
dispatched messages but since you are still connected you could miss sending acks.  I'd advise
rolling that back.  

If there's any failure recovery needed then we should be doing down at the failover transport
level where the state tracker and connection interruption processing is properly handled.
 I think the change may also cause mismatched interrupted / resumed calls as well.  

If you have some broker and client logs from the exception you are seeing I'd be happy to
look at them and see what's going on.
                
> NMSConnectionException does not trigger failover recovery.
> ----------------------------------------------------------
>
>                 Key: AMQNET-371
>                 URL: https://issues.apache.org/jira/browse/AMQNET-371
>             Project: ActiveMQ .Net
>          Issue Type: Bug
>          Components: ActiveMQ, Stomp
>    Affects Versions: 1.5.3
>            Reporter: Jim Gomes
>            Assignee: Jim Gomes
>              Labels: broker, error, exception, failover
>             Fix For: 1.5.4, 1.6.0
>
>   Original Estimate: 48h
>  Remaining Estimate: 48h
>
> When a failover protocol is used, it is expected that any network interruptions will
be automatically recovered.  Currently, an internal broker error send an error message to
a client and kick it off.  The client then throws an NMSConnectionException.  This exception
is not automatically handled by the failover protocol, and the client application is then
forced to completely destroy and rebuild all connections, sessions, consumers and producers,
if that's even possible without restarting.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message