activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ron Koerner (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AMQ-4097) Broker-to-Broker Reconnect fails wrongly due to duplicate name
Date Thu, 21 Nov 2013 10:54:36 GMT

    [ https://issues.apache.org/jira/browse/AMQ-4097?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13828815#comment-13828815
] 

Ron Koerner commented on AMQ-4097:
----------------------------------

Hi Amine,

usually that meant, that the broker has a lot of things to do. Especially broker logins/logouts
seem to be quite critical. We have one environment with about 50 connected remote brokers
which have about 10000 consumers shared among them. These are all connected over one leased
line, so when that line breaks down, all remote brokers are disconnected due to missing keepalive.
That causes a cascade of teardowns which can take minutes, especially if the line comes back
and the remote brokers try to login again.

We do not have problems with a high number of messages to be sent in either direction or one
logout/login.

Regards,
Ron

> Broker-to-Broker Reconnect fails wrongly due to duplicate name
> --------------------------------------------------------------
>
>                 Key: AMQ-4097
>                 URL: https://issues.apache.org/jira/browse/AMQ-4097
>             Project: ActiveMQ
>          Issue Type: Bug
>    Affects Versions: 5.7.0
>         Environment: A central broker to which a lot (50+) of external brokers connect
with a duplex bridge. A special routing/firewall is used which can affect timing but not order
of TCP packets. This can be simulated by using socat.
> Actually we are using 5.7-SNAPSHOT of 2012-08-31.
>            Reporter: Ron Koerner
>
> The situation is as follows:
> - an external broker A connects
> - time passes
> - a lot of external brokers disconnect including A
> - A reconnects (as well as all the other external brokers)
> - wrong message about duplicate name is generated
> In the log it looks like this:
> {code}
> 2012-10-08 17:11:19,835 INFO  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#278 and tcp:///127.0.0.1:54191(cbox-56BU902442) has been established. [StartLocalBridge:
localBroker=vm://c04ptec#278]
> ...
> ... a lot more of the following with different ports
> 2012-10-08 17:37:01,958 WARN  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#278 and tcp:///127.0.0.1:54191 shutdown due to a remote error: java.io.EOFException
[ActiveMQ NIO Worker 193]
> ... more of these
> 2012-10-08 17:37:03,438 INFO  emq.broker.TransportConnection - Started responder end
of duplex bridge cBox 56BU902442 to cBox Proxy@ID:P013SPWMK1WN-39320-1349704902319-0:1 [ActiveMQ
NIO Worker 215]
> ...
> 2012-10-08 17:37:03,922 WARN  emq.broker.TransportConnection - Failed to add Connection
ID:c04ptec-51799-1349706422094-242:2, reason: javax.jms.InvalidClientIDException: Broker:
c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_inbound_c04ptec already connected
from vm://c04ptec#278 [StartLocalBridge: localBroker=vm://c04ptec#478]
> 2012-10-08 17:37:03,923 INFO  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#478 and tcp:///127.0.0.1:56529 shutdown due to a local error: javax.jms.InvalidClientIDException:
Broker: c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_inbound_c04ptec already
connected from vm://c04ptec#278 [StartLocalBridge: localBroker=vm://c04ptec#478]
> ...
> 2012-10-08 17:37:04,036 INFO  .DemandForwardingBridgeSupport - c04ptec bridge to cbox-56BU902442
stopped [ActiveMQ Task-182]
> ...
> 2012-10-08 17:37:06,540 INFO  emq.broker.TransportConnection - Started responder end
of duplex bridge cBox 56BU902442 to cBox Proxy@ID:P013SPWMK1WN-39320-1349704902319-0:1 [ActiveMQ
NIO Worker 207]
> ...
> 2012-10-08 17:37:06,548 WARN  emq.broker.TransportConnection - Failed to add Connection
ID:c04ptec-51799-1349706422094-292:1, reason: javax.jms.InvalidClientIDException: Broker:
c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_inbound_c04ptec already connected
from vm://c04ptec#278 [StartLocalBridge: localBroker=vm://c04ptec#570]
> 2012-10-08 17:37:06,548 INFO  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#570 and tcp:///127.0.0.1:56576 shutdown due to a local error: javax.jms.InvalidClientIDException:
Broker: c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_inbound_c04ptec already
connected from vm://c04ptec#278 [StartLocalBridge: localBroker=vm://c04ptec#570]
> ...
> 2012-10-08 17:37:06,559 INFO  .DemandForwardingBridgeSupport - c04ptec bridge to cbox-56BU902442
stopped [ActiveMQ Task-204]
> ...
> 2012-10-08 17:37:24,417 INFO  .DemandForwardingBridgeSupport - c04ptec bridge to cbox-56BU902442
stopped [ActiveMQ Task-73]
> ...
> 2012-10-08 17:37:25,103 INFO  emq.broker.TransportConnection - Started responder end
of duplex bridge cBox 56BU902442 to cBox Proxy@ID:P013SPWMK1WN-39320-1349704902319-0:1 [ActiveMQ
NIO Worker 268]
> ...
> 2012-10-08 17:37:29,110 INFO  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#594 and tcp:///127.0.0.1:56656(cbox-56BU902442) has been established. [StartLocalBridge:
localBroker=vm://c04ptec#594]
> ...
> 2012-10-08 17:37:59,669 WARN  .DemandForwardingBridgeSupport - Network connection between
vm://c04ptec#594 and tcp:///127.0.0.1:56656(cbox-56BU902442) was interrupted during establishment.
[StartLocalBridge: localBroker=vm://c04ptec#594]
> ...
> 2012-10-08 17:38:09,005 INFO  .DemandForwardingBridgeSupport - c04ptec bridge to cbox-56BU902442
stopped [ActiveMQ Task-228]
> ...
> 2012-10-08 17:38:18,681 INFO  emq.broker.TransportConnection - Started responder end
of duplex bridge cBox 56BU902442 to cBox Proxy@ID:P013SPWMK1WN-39320-1349704902319-0:1 [ActiveMQ
NIO Worker 292]
> 2012-10-08 17:38:18,681 WARN  emq.broker.TransportConnection - Failed to add Connection
ID:P013SPWMK1WN-39320-1349704902319-152:1, reason: javax.jms.InvalidClientIDException: Broker:
c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_outbound already connected
from vm://c04ptec#594 [ActiveMQ NIO Worker 292]
> 2012-10-08 17:38:18,682 WARN  er.TransportConnection.Service - Async error occurred:
javax.jms.InvalidClientIDException: Broker: c04ptec - Client: cBox 56BU902442 to cBox Proxy_cbox-56BU902442_outbound
already connected from vm://c04ptec#594 [ActiveMQ NIO Worker 292]
> javax.jms.InvalidClientIDException: Broker: c04ptec - Client: cBox 56BU902442 to cBox
Proxy_cbox-56BU902442_outbound already connected from vm://c04ptec#594
>         at org.apache.activemq.broker.region.RegionBroker.addConnection(RegionBroker.java:222)
~[activemq-core-5.7-SNAPSHOT.jar:5.7-SNAPSHOT]
> {code}
> This repeats until the external broker is stopped and started manually.
> I assume there is a race condition which shows up under high load preventing the information
about the disconnected client to be removed before the same client reconnects.



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message