Hi all.
I'm trying for two days now, getting HADB working with high availability session failover.
The problem is: the session simply doesn't get stored in the db.
I'm using GF V2 rc1 (b58) and installed the hadb implementation from sun which are available
at Sun.com if you download the Sun Enterprise System (java_es-5-ga-linux-x86.zip).
For getting the enterprise profile working I had to install the NSS-rpms fom the same zip and change some parameters in the asenv.conf in glassfish/config directory.
My environment constis of two clustered machines (one Intel Core 2 and one double-Xeon) with plenty of RAM (4G & 2G) and a domain administration server.
I set up a HADB instance with two nodes (co-located on the cluster machines) which was no problem. The jdbc connection is working and the cluster nodes talk with each other.
The setup was done with asadmin configure-ha-cluster without problems.
I enabled everything's necessary, from <distributable> tag in web.xml, over availability settings in the cluster configuration to web container availability (type=ha, frequency=web-method, scope=session).
I'm using jk_mod for load balancing.
Yet I get no failover if I withdraw one instance from the load balancer and the request switches over to another node. The session is not in the database (verified with clusql from hadb rpm) and I'm getting a new session ID everytime a switch takes place.
Any ideas?
Please help!
Alexander.
[Message sent by forum member 'sahlix' (sahlix)]
http://forums.java.net/jive/thread.jspa?messageID=230239