From artpoon at gmail.com  Tue Mar 30 11:25:34 2010
From: artpoon at gmail.com (Art Poon)
Date: Tue Nov  9 01:14:29 2010
Subject: [scyld-users] Re: spawning remote java processes
Message-ID: <8D1E6A9B-FD73-4C08-AB5D-21AB05EF32FD@gmail.com>

Hi there,

I'm running a Scyld cluster with CentOS 5.4.  Lately I've been trying to configure the cluster to allow me to run Java processes remotely (I don't program in Java myself but need this particular program).  

Initially I ran into the following with 'bpsh 0 java':

> java: error while loading shared libraries: libjli.so: cannot open shared object file: No such file or directory 

which I resolved (with help from Penguin support) by adding the following to /etc/exports:

> /usr/share/java @cluster(ro,sync,no_root_squash)
> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/bin @cluster(ro,sync,no_root_squash)
> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib @cluster(ro,sync,no_root_squash)
> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64 @cluster(ro,sync,no_root_squash)
> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64/jli @cluster(ro,sync,no_root_squash)

and to /etc/beowulf/fstab:

> 
> $MASTER:/usr/share/java		/usr/share/java		nfs	nolock,nonfatal0 0
> $MASTER:/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/bin		/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/bin		nfs	nolock,nonfatal0 0
> $MASTER:/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib		/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib		nfs	nolock,nonfatal0 0
> $MASTER:/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64	/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64	nfs	nolock,nonfatal0 0
> $MASTER:/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64/jli	/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64/jli		nfs	nolock,nonfatal		0 0

This resolved the missing 'libjli.so' issue, but now I get the following errors:

> Error: could not find libjava.so
> Error: could not find Java 2 Runtime Environment.

which is odd because the path to libjava.so should be mounted:

> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/lib/amd64/libjava.so

and I have this in my LD_LIBRARY_PATH.  I'm sure I'm overlooking something silly, any help much appreciated!

Thanks,
- Art.