At a client the Oracle Service Bus is running on the Windows platform. We had some strange problems with the Weblogic Managed Servers shutting down suddenly without any messages handled by the OSB at the moment.
The managed server logging only showed the following line:
####<someDateTime> <Notice> <WebLogicServer> <someServer> <someManagedServer> <Thread-1> <<WLS Kernel>> <> <> <1280144706113> <BEA-000388> <JVM called WLS shutdown hook. The server will force shutdown now>
At first we couldn’t pinpoint the problem, but eventually we found this post on the Oracle Forums.
quote from post by Umesh Kumar Maurya:
What we have found in our environment that the managed server goes down whenever administrator user logs into Windows server using remote desktop in console mode (e.g. mstsc /console) and then logs off. If the administrator logs into the server in regular mode (e.g. mstsc) and then logs off, this problem is not simulated. To resolve this problem, you can pass -Xnohup parameter to JVM.
We checked our Windows Security Logs (computer management -> System Tools -> Event Viewer -> Security) and indeed discovered that at the same time we did see the JVM WLS shutdown hook in our logging, someone from system engineering connected to the Windows console with RDP.
Adding the Xnohup parameter to the JVM fixed the problem for us.
Jacco is correct. JRockit now a days also supports -Xrs like Sun’s Hotspot JVM. So I guess it’s better to just use Xrs since it always works and doesn’t break when you switch JDK from JRockit to Sun.