1_4_2_05 VM's crashing unexpectedly on RedHat AS3.0

Question

Hi. I'm new to this forum so forgive me for any bad posting practices.
We are running java 1.4.2_05 on RedHat Advance Server 3.0 2.4.21-20.ELsmp
Over the last few weeks ( and possibly longer) individual VMs seem to be falling over in circumstances we do not understand. They may be related, or they may not.
There seem to be two different errors happening.
1) Virtual machine exited with code 129.
The VM exits with code 129. We think this may indicate an external signal to the VM, but cannot account for that. This is often seen following a long Full GC. Normally these last 10-20 seconds, but when this error is seen the GC may have lasted many hundreds of seconds. This could indicate some problem with the host systems (Xeon 32-bit dual CPU, 6GB RAM) but here are no messages in the OS logs to indicate that Linux is killing the vm.
2) Exception java.lang.OutOfMemoryError: requested 1000 bytes for promotion. Out of swap space?
This error seems to indicate that the OS is unable to provide memory. I have seen on the web some references to errors like this, but these seem to suggest that you can work round this error by pre-allocating the Java memory by setting the -ms parameter to the same value as -mx. They also suggest upping the max perm size and changing the object evacuation ratio to modify the GC behaviour but these seem to make no difference either.This has been done, but the error persists. Again there seems to be no resource issues on the LINUX server associated with this event.
full parameters are:
[/opt/DKBjava/1_4_2_05/jre/bin/java, -server, -Xms2000M, -Xmx2000M, -verbose:gc, -XX:+UseParNewGC, -XX:+UseAdaptiveSizePolicy, -Dsun.rmi.dgc.client.gcInterval=72000000, -Dsun.rmi.dgc.server.gcInterval=72000000,  -XX:MaxLiveObjectEvacuationRatio=50, -Dsun.rmi.server.exceptionTrace=true
we also run
[/opt/DKBjava/1_4_2_05/jre/bin/java, -server, -Xms2300M, -Xmx2300M, -verbose:gc, -XX:+UseParNewGC, -XX:+UseAdaptiveSizePolicy, -Dsun.rmi.dgc.client.gcInterval=72000000, -Dsun.rmi.dgc.server.gcInterval=72000000, XX:MaxPermSize=256M, -XX:MaxLiveObjectEvacuationRatio=50, -Dsun.rmi.server.exceptionTrace=true,
on some of the other vms as well but these fail as well...
Any ideas, greatly appreciated.

Answer 1

Thanks for the speedy reply.
I believe there used to be an upper limit on memory size of around 2GB, but with AS3.0 this is larger.
The overall amount of heap that you can actually allocate at runtime has gone up from ~1.9GB to 2.65GB on AS3.0
You used to get this message when you exceeded the permissible max:
"Error occurred during initialization of VM
Could not reserve enough space for object heap"
My belief is that this statement refers to that, can anyone confirm this ?