ActiveMQ and failover

[Danielius Jurna <da...@elitnet.lt>]

It seems, that ActiveMQ 4.0 final has been tested very little to handle
failover scenarios, because simple cable disconnects brings the broker down.
I've posted those problems in the forum before, but seems nobody wants to
answer (or don't know the answer).
http://www.nabble.com/ActiveMQ+broker+brokes+down+on+client+reconnect-t1682129.html
http://www.nabble.com/Stomp+subscribtions+are+not+removed-t1627261.html
I know, that it's free software and I beleive in ActiveMQ very much, but
currenlty version 4.0 has very childish problems and 3.2 has some issues
which will not be corrected because of desingn flaws (after 100000
unconsumed messages, 3.2 servers brokes down whithout any chance to
recover).

Please, take a look at those failover problems!
--
View this message in context: http://www.nabble.com/ActiveMQ+and+failover-t1688385.html#a4581475
Sent from the ActiveMQ - User forum at Nabble.com.

Re: ActiveMQ and failover

[Danielius Jurna <da...@elitnet.lt>]

I think the error handling in AbstractConnection has some design flaws.
Before performing any operation on the broker (addConsumer, addDestination),
the state of the conenction is checked. But during that operation async
error can occure (write error, read error and seems registerMBean exception)
and connection is removed. But in AbstractConnection this condition is not
checked and consumer or producer is added to SessionState and broker.
In unreliable network queues and Subscriptions sometimes become
non-functional. Sometimes it is not possible to subscribe to existing queue
(cannot register MBean exceptions are most often seen - but I think this is
a consequence, not the cause). And you need to restart the broker.
--
View this message in context: http://www.nabble.com/ActiveMQ+and+failover-t1688385.html#a4610431
Sent from the ActiveMQ - User forum at Nabble.com.

Re: ActiveMQ and failover

[Danielius Jurna <da...@elitnet.lt>]

I've added issue considering Stomp client in jira:
http://issues.apache.org/activemq/browse/AMQ-724
And I've aded a test case which reproduces this issue, and quick soluton for
that. But this bug needs real fix.
--
View this message in context: http://www.nabble.com/ActiveMQ+and+failover-t1688385.html#a4606302
Sent from the ActiveMQ - User forum at Nabble.com.

Re: ActiveMQ and failover

[amerigo5 <am...@hotmail.com>]

This is one of the scenarios on my list to test.  I think this is a bug.  I
hope this gets a necessary attention soon.  Thanks.
--
View this message in context: http://www.nabble.com/ActiveMQ+and+failover-t1688385.html#a4582241
Sent from the ActiveMQ - User forum at Nabble.com.

Re: ActiveMQ and failover

[Janet Cooper <fi...@gmxpro.net>]

I think I may have the same problem. 

Setup:
  2 Brokers configured as a network of brokers (jdk 1.5.0_06 / linux
(ubuntu))
  
Reproduce:
1. Start Broker A
2. Start Broker B
3. Wait until Broker B is started. Kill Broker B
4. Restart Broker B (not necessary since Broker A can't connect anyway)

Output from Broker A:
java.io.EOFException
	at java.io.DataInputStream.readInt(DataInputStream.java:358)
	at
org.apache.activemq.openwire.OpenWireFormat.unmarshal(OpenWireFormat.java:270)
	at
org.apache.activemq.transport.tcp.TcpTransport.run(TcpTransport.java:142)
	at java.lang.Thread.run(Thread.java:595)
03:39:04,664  WARN DemandForwardingBridge:133 - Outbound transport to
localhost interrupted ...
03:40:20,842  INFO DemandForwardingBridge:491 - localhost Shutting down
03:40:20,852 DEBUG DemandForwardingBridge:251 -  stopping localhost bridge
to localhost is disposed already ? false
03:40:20,862 DEBUG FailoverTransport:251 - Stopped.
03:40:20,874 DEBUG DemandForwardingBridge:276 - localhost bridge to
localhost stopped
03:41:28,224 DEBUG FailoverTransport:226 - Transport failed, starting up
reconnect task
java.io.EOFException
	at java.io.DataInputStream.readInt(DataInputStream.java:358)
	at
org.apache.activemq.openwire.OpenWireFormat.unmarshal(OpenWireFormat.java:270)
	at
org.apache.activemq.transport.tcp.TcpTransport.run(TcpTransport.java:142)
	at java.lang.Thread.run(Thread.java:595)

...and then silence. It never reconnects.
--
View this message in context: http://www.nabble.com/ActiveMQ+and+failover-t1688385.html#a4595884
Sent from the ActiveMQ - User forum at Nabble.com.