5.2.2 Clustered Ehcache problem?

[Jesper W, modificado 15 Anos atrás.]

This morning we attempted to deploy a 5.2.2 based release to our production cluster.
It failed. Unfortunatly we have only a single VM stageing environment to test in... ;-)

Has anyone successfully started Liferay 5.2.2 clustered?

It seems to me the problem would show everywhere.

The traceback of the problem (IP adresses censored with ....):

java.lang.ExceptionInInitializerError
	at com.liferay.portal.events.StartupAction.run(StartupAction.java:84)
....
Caused by: net.sf.ehcache.CacheException: Problem starting listener for RMICachePeer //...:53270/com.liferay.portal.kernel.dao.orm.FinderCache#com.liferay.portal.model.Company. Initial cause was invalid character, '#', in URL name: //...:53270/com.liferay.portal.kernel.dao.orm.FinderCache#com.liferay.portal.model.Company
	at java.lang.Thread.run(Thread.java:619)
	at net.sf.ehcache.distribution.RMICacheManagerPeerListener.notifyCacheAdded(RMICacheManagerPeerListener.java:513)
....
	at com.liferay.portal.cache.EhcachePortalCacheManager.getCache(EhcachePortalCacheManager.java:75)
....
Caused by: java.net.MalformedURLException: invalid character, '#', in URL name: //...:53270/com.liferay.portal.kernel.dao.orm.FinderCache#com.liferay.portal.model.Company
	at java.rmi.Naming.intParseURL(Naming.java:262)
	at java.rmi.Naming.parseURL(Naming.java:220)
	at java.rmi.Naming.rebind(Naming.java:154)
	at net.sf.ehcache.distribution.RMICacheManagerPeerListener.bind(RMICacheManagerPeerListener.java:230)

The same server runs the 5.1 release just fine.
The #-character in the cache name has been there since 5.1.
So compared to 5.1 we have the same URL, the same Java, but an updated Ehcache version.
Not shure what to make of that...

Any ideas?

[Mika Koivisto, modificado 15 Anos atrás.]

EhCache was upgraded to 1.5 so could that cause the problem. I'm not familiar with EhCache RMI inner workings but I'm just wondering why is the hash mark part of the rmiUrl shouldn't that be the cache key.

[Jesper W, modificado 15 Anos atrás.]

Actually the same problem occurs if I replace with echache.jar 1.4.1
So it seems to be Liferay related.

It's easy to reproduce, just take 5.2.x branch and change to:

net.sf.ehcache.configurationResourceName=/ehcache/hibernate-clustered.xml
ehcache.multi.vm.config.location=/ehcache/liferay-multi-vm-clustered.xml

Start server and watch the crash...

[Jesper W, modificado 15 Anos atrás.]

Also verified that the same problem exists in trunk.

[Jonas Yuan, modificado 15 Anos atrás.]

Hi Jesper,

Thank you!

According to our experience in Sesame Street, the clustering settings must be used in the clustering environment only.

That is, in the normal environment, we have to remove the clustering settings. Only in the clustering environment, we need to turn on the clustering settings. Does this make sense?

Best regards,

Jonas Yuan

[Jesper W, modificado 15 Anos atrás.]

It makes sense, but I'm afraid exactly the same crash happens in the real clustered environment.
Have you gotten 5.2.2 running clustered?

[Brian Chan, modificado 15 Anos atrás.]

Fixed in LPS-2120.

[Jonas Yuan, modificado 15 Anos atrás.]

Thank you so much, Brian!

It was getting fixed so quick ....

[Jesper W, modificado 15 Anos atrás.]

For the record, there was actually two problems here. Fixing the first one exposed a class loader problem in JBoss which is now fixed in LPS-2178.
/j

[Marco Altieri, modificado 15 Anos atrás.]

Hi all,
where can I get the a patch for this bug?

thanks

[Jesper W, modificado 15 Anos atrás.]

Read the respective LPS issues on http://support.liferay.com.

[Marco Altieri, modificado 15 Anos atrás.]

Ok, thank you!
Marco