Pure Master Slave

2012-01-04 Thread scottyob
Hi all. I'm trying to follow the example at http://activemq.apache.org/pure-master-slave.html to set up a pure master slave. I'm a bit confused about the brokers in the xml file. Do I add a 2nd in the example config that comes bundled with ActiveMQ or replace the ones? It lists two examples of

Pure Master/Slave. Slave keeps the messages already deleted from Master

2012-01-04 Thread SergueiM
Hi, Setup a Pure Master/Slave with KahaDB. After deleting a message previously sent to the Master, I am still seeing it in Slave. The expectation was that the message would disappear once consumed/deleted. Using web admin console: (1) submitted a non-persistent message to Master; (2) a copy showe

Re: Virtual topic subscriber expiration?

2012-01-04 Thread Jason Dillon
On Jan 4, 2012, at 4:36 AM, Dejan Bosanac wrote: > There's no explicit setup for that as those are just regular queues from > which no one consumes. The simple manual solution is to just delete unused > queues and virtual topic will stop sending messages there > > Some things you can do to automat

php ack problem

2012-01-04 Thread Christopher Wood
I have a problem with ActiveMQ not removing messages from a queue when the ack is received from PHP. Do you list denizens possibly know what's going on? As background, it appears that Net::STOMP::Client adds a content-length header to a message, whereas PHP's Stomp doesn't. When Net::STOMP::Clie

should I use trunk or patch 5.5.1?

2012-01-04 Thread mserrano
Hi, I need the fix for https://issues.apache.org/jira/browse/AMQ-3542. Should I try to patch 5.5.1 up as needed to include the fixes for it? Or should I use trunk? I'm not sure how stable trunk is or how close 5.6 is to shipping. Thanks, Martin p.s. I'm trying to deal with a situation where

Checking that slave is up

2012-01-04 Thread Stevo Slavić
Hello ActiveMQ community, Is there a way to check that a slave, in shared filesystem storage master/slave setup which is awaiting lock to become available, is up? Does JMX interface start early enough, before lock is acquired? Regards, Stevo.

Re: nio+ssl w/programatic {key|trust}store configuration

2012-01-04 Thread Jason Dillon
On Jan 4, 2012, at 5:41 AM, Dejan Bosanac wrote: > nio+ssl should work exactly the same as the plain ssl transport, so this > SslBrokerService enhancement you're working is totally valid and we should > have it. Okay I think I can make a simple test to show working ssl and non-working nio+ssl the

Re: broker dataDirectory and kahaDB directory

2012-01-04 Thread Stevo Slavić
Dejan, You stated "By default kahaDB uses dataDir/kahadb". I guess by dataDir you mean here on broker dataDirectory attribute value. Problem is, docs found here http://activemq.apache.org/kahadb.html state that different value, "activemq-data", is being used as kahaDB element directory attribute d

Re: JMSXGroupID timeout

2012-01-04 Thread Dejan Bosanac
There is no one-to-one mapping between groups and consumers, it's just that messages from one group are guaranteed to be processed by a single consumer. Regards -- Dejan Bosanac - http://twitter.com/dejanb - The experts in open source integration and messaging - http://fusesource.

Database locked wait delay longer than databaseLockedWaitDelay

2012-01-04 Thread Stevo Slavić
Hello ActiveMQ community, While testing shared fileystem storage master/slave setup behavior when master node loses access to shared filesystem storage and gets fenced, on slave node databaseLockedWaitDelay unexpectedly did not get respected, actual delay was much longer at a single last check. Lo

Re: nio+ssl w/programatic {key|trust}store configuration

2012-01-04 Thread Dejan Bosanac
nio+ssl should work exactly the same as the plain ssl transport, so this SslBrokerService enhancement you're working is totally valid and we should have it. The reason why socket factory isn't used in nio case is because it isn't used by the java nio ssl API (if you hate ssl you'll hate nio+ssl ev

Re: JMSXGroupID timeout

2012-01-04 Thread pwanner
It seems that I was wrong thinking that there will be one consumer for every groupId, but only a mapping between the groupId hash and the max available consumers, thus there is no real need for cleaning the mapping groupId-to-consumer. http://stackoverflow.com/questions/703659/activemq-message-gro

Re: broker dataDirectory and kahaDB directory

2012-01-04 Thread Dejan Bosanac
Yes, log4j ones Regards -- Dejan Bosanac - http://twitter.com/dejanb - The experts in open source integration and messaging - http://fusesource.com ActiveMQ in Action - http://www.manning.com/snyder/ Blog - http://www.nighttale.net 2012/1/4 Stevo Slavić > OK, thanks! I suppos

Re: broker dataDirectory and kahaDB directory

2012-01-04 Thread Stevo Slavić
OK, thanks! I suppose you mean on ActiveMQ application log files (each node to have it's own), and not on journal logs. Regards, Stevo. 2012/1/4 Dejan Bosanac > General good practice is to keep them set to the same dir as brokers can > share scheduler info as well (if that feature is used). How

Re: Virtual topic subscriber expiration?

2012-01-04 Thread Dejan Bosanac
There's no explicit setup for that as those are just regular queues from which no one consumes. The simple manual solution is to just delete unused queues and virtual topic will stop sending messages there Some things you can do to automate this process is: - set expiry on the messages you're sen

Re: broker dataDirectory and kahaDB directory

2012-01-04 Thread Dejan Bosanac
General good practice is to keep them set to the same dir as brokers can share scheduler info as well (if that feature is used). However you might consider keeping logs locally. Regards -- Dejan Bosanac - http://twitter.com/dejanb - The experts in open source integration and mess