[clebert.suconic]

"timfox" wrote:
Clebert-

I just tried to deploy the scoped sar in JBAS and it fails, complaining that it cannot find a property editor for ConnectionFactory::setLoadBalancingFactory.

Looking at the code, the method setLoadBalancingFactory() takes a LoadBalancingFactory, but the MBean method takes a String.

I have fixed this, so I can get on with my work. You may want to review my fix since it was very quick.

This tells me one thing, you added an MBean attribute without adding a corresponding test for it.

All new functionality needs to have unit tests, so please could you add a test.

Cheers.

I just changed this in top of what Ovidiu had before... Ovidiu had a LoadBalancingPolicy and I placed a LoadBalancingFactory in replacement.

I actually started doing this by using a String the same way you changed... but then I realized what Ovidiu had before and I used the same thing. By looking at his code I thought MBeans would take care of it on the XML description (using a default's constructor)

[timfox]

The update is happening on ConnectionFactoryJNDIMapper::onReplicationChange..


Look at the method's implementation, you will see a call to updateClusteredClients:

line 359: endpoint.updateClusteredClients(delArr, failoverMap);

Sigh... You're still missing the point :)

This line of code is only called when the replication change is a deploy/undeploy of a connection factory:

else if (sKey.startsWith(CF_PREFIX) && originatorNodeID != serverPeer.getServerPeerID())
 {

My question was, can you explain to me how this is called when a node crashes?

When a node crashes it won't have the courtesy to undeploy its connection factories first.

So how would this get called?

[clebert.suconic]

Sigh... You're still missing the point :)

This line of code is only called when the replication change is a deploy/undeploy of a connection factory:

Nope... this place is called also when a node leaves/joins the cluster. Replicator is re-sending the lists when a node leaves... (crash).

This is exactly the place where we were doing rebinds before.


I just executed ClusterViewUpdate Test, using run test... and I could see the event fired on logs.

ClusterViewUpdate would fail if the Connection Factory was not updated after a crash.

[timfox]

line of code please?

[timfox]

CF_PREFIX is used when a connection factory is deployed/undeployed.

ADDRESS_INFO_KEY is used when a node joins/leaves

What you are now saying is that we make a change with prefix CF_PREFIX when a node joins leaves?

[timfox]

Looking through the code, the only place where a replication change with prefix CF_PREFIX = "CF_" is sent is when a connection factory is deployed or undeployed, i.e. in registerConnectionFactory() and unregisterConnectionFactory().

There is no where else in the code where such a replication change is propagated.

So you still haven't explained how such a replication event gets propagated when a node crashes.....

[clebert.suconic]

- DefaultClusteredPostOffice::nodeLeft will call cleanLocalDataForNode.

cleanLocalDataForNode will resubmit all the listeners as the CF was being redeployed. This logic was written long time ago.. I didn't change anything on the rebinds logics.