SEVERE: The web application [/cas] registered the JBDC driver [com.mysql.jdbc.Driver] but failed to unregister it when the web application was stopped. To prevent a memory leak, the JDBC Driver has been forcibly unregistered.
This is a warning emitted by the Tomcat memory leak detection components. See http://wiki.apache.org/tomcat/MemoryLeakProtection for more information. We have not audited CAS to address all the cases on that page, but the most common cases are application-spawned thread pools as with the Quartz timers and thread-local classloader leaks. In any case we don't really support hot deploys of CAS on Tomcat, or any other container, for that matter. To be safe you should always restart your servlet container on redeployment. There are a number of HA setups that will provide robust and graceful failover during servlet container restarts if you should need a very highly-available CAS architecture. We can provide more information if needed.
I did noticed that after few stops/starts tomcat memory allocation is growing high and the cpu of the machine is raising up to 99% of usage.
The log entries you cited are from a running Tomcat and emanate only when servlets are undeployed from the running container, which is to say it has nothing to to with excessive memory and CPU once Tomcat has stopped. The sense of "leak" in the log entries is a resource leak within the JVM, which would disappear as soon as the JVM is terminated.
M -- You are currently subscribed to [email protected] as: [email protected] To unsubscribe, change settings or access archives, see http://www.ja-sig.org/wiki/display/JSG/cas-user
