activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From contactreji <contactr...@gmail.com>
Subject Active MQ Clustering - Lock file issue
Date Fri, 29 Nov 2013 07:01:15 GMT
Hi 

I am trying to implement high availability feature  by clustering instances
of Active MQ.

I had initially set up 3 instances with single shared state by keeping a
common file drive as

   <persistenceAdapter>
            <kahaDB directory="F:/Data"/>
    </persistenceAdapter>

This was entered in conf file of all AMQ instances.

Its working absolutely fine. When one broker goes down gracefully, the next
one which is in blocked state due to lock acquired by the first one indeed
comes up and takes over as master.

But, when i abruptly disconnect one broker (by disrupting network
connections etc), the lock file is still held on the shared resource and the
backup brokers are not able to come up for 120 seconds.

Can you help me in configuring the slaves to hear the heart beat of the
master instance. 

1) Is there a provision that we can force the slave instances to immediately
take over the shared states when the heart beat for previous master is gone?
2) Is there a provision that we can force the Operating System to clear the
lock file as soon as the master gets disconnected due to some catastrophic
failure?

Any ideas will be really helpful!

Reji





--
View this message in context: http://activemq.2283324.n4.nabble.com/Active-MQ-Clustering-Lock-file-issue-tp4674937.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Mime
View raw message