running livecycle es 8.2 on suse linux enterprise server
Hi, I've recently received a strange bug while trying to startup jboss livecycle on suse linux enterprise server. The error that occurs first is:<br /><br />2008-10-23 10:58:23,065 ERROR [org.jboss.deployment.MainDeployer] Could not initialise deployment: file:/app/jboss/server/all/deploy/adobe-livecycle-native-jboss-x86_linux.ear<br />java.lang.OutOfMemoryError<br /> at java.util.zip.ZipFile.open(Native Method)<br /> at java.util.zip.ZipFile.<init>(ZipFile.java:203)<br /> at java.util.jar.JarFile.<init>(JarFile.java:132)<br /> at java.util.jar.JarFile.<init>(JarFile.java:70)<br /><br />After some research i found out that this is not a java heap space bug nor a livecycle bug , but a java 1.5 bug, the bug database url is as follows:<br /><br />http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4705373<br />duplicate of:<br />http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6280693<br /><br />As stated this is a java bug that was fixed in the mustang release (jdk1.6), but since we cant use jdk1.6, is it even possible to deploy lc on suse linux enterprise server ?
You usually get java.lang.OutOfMemoryError if you have given your JVM not enough memory (heap). If that is the case with you, this is not a bug at all. It does not matter if your box has 8 GB of RAM if your JVM heap settings are set to maximum 256 MB!
Please see this and make changes to your JBoss run.sh
Hi Jayan, thanks for the reply, but as I stated, this is not a java heap space problem. I have done all the necesary tests to confirm that, the XmX is set to 1280m which is the maximum for the 32-bit jdk and at the point of the crash there's still plety of java heap space avaible (the crash occurs after like a 3 minuts already in the early stage of the deployment phae). Also, when Livecycle does run out of heap space (which happened to us on some windows machines), you get a clear message java.lang.OutOfMemory : Out of heap space, which does not occur in our case.
So to clariy, the java heap space is here not the problem, the bug is the same as in sun's bug database (see links for details) and has to do with the mmap system call under linux.