I am working on enhancing the cluster support in APLAWS+. We have an HTTP load balancer out the front of our Tomcats and have noticed a couple of deficiencies.
The most critical one for us right now is that if a user-defined content type is created or changed in one Tomcat, the other Tomcat throws an exception if that content type is referenced because the ObjectType isn't found or is stale. The other Tomcat(s) needs to (re)load the ObjectType whenever content types are changed. At the moment I am using ActiveMQ to publish notifications via JMS. ActiveMQ has embedded JMS brokers with multicast to auto-detect other nodes in the cluster so there are no deployment changes to APLAWS.
I will also look at using ActiveMQ rather than HTTP for CacheTable synchronisation, possibly converting it to http://ehcache.sourceforge.net/. One issue with CacheTable is that purging the entire table isn't propagated even with HTTP.
Any comments would be appreciated.
Regards,
Terry
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi, Not used jms for clustering. I understand camden are running aplaws out of jboss and this is likely using the jboss clustering features. Terry what sites are you running APLAWS on?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am working on enhancing the cluster support in APLAWS+. We have an HTTP load balancer out the front of our Tomcats and have noticed a couple of deficiencies.
The most critical one for us right now is that if a user-defined content type is created or changed in one Tomcat, the other Tomcat throws an exception if that content type is referenced because the ObjectType isn't found or is stale. The other Tomcat(s) needs to (re)load the ObjectType whenever content types are changed. At the moment I am using ActiveMQ to publish notifications via JMS. ActiveMQ has embedded JMS brokers with multicast to auto-detect other nodes in the cluster so there are no deployment changes to APLAWS.
I will also look at using ActiveMQ rather than HTTP for CacheTable synchronisation, possibly converting it to http://ehcache.sourceforge.net/. One issue with CacheTable is that purging the entire table isn't propagated even with HTTP.
Any comments would be appreciated.
Regards,
Terry
Hi, Not used jms for clustering. I understand camden are running aplaws out of jboss and this is likely using the jboss clustering features. Terry what sites are you running APLAWS on?