Submitting a job from computing nodes other than a head node

[Hamid3252]

Hi everybody,

We installed ABAQUS on one of our clusters, and the server has two nodes, each having one Dual-core CPU and 2GB RAM. When we log on to the system, we are on the head node which is the default one. To change the node, we write "ssh compute-0-nodeNumber", but when we change the computing node, we don't have access to ABAQUS files anymore, because when I log on to the a computing node (e.g. "ssh compute 0-1"), ABAQUS path is not defined in PATH variable, and neither can I find it on the specified path which is seen from the head node (/opt/ABAQUS/etc...). How can I access that path and what is the best way to use those nodes to run a job?

Thank you,

[avatar]

What is the exact error message you are getting when you are trying to access it?

[lkewei]

You cannot run Abaqus in this way, you have to install Abaqus in a shared directory to run it in compute nodes. Becauce your Abaqus has only been installed on the head node. Once you log on to compute node, the Abaqus files in /opt/ABAQUS/ on head node are not accessable from compute nodes. you can do either one of these:
1. install abaqus on compute node, since you only have two nodes. But if you reinstall your computes nodes, you have to reinstall abaqus agian.
2. install abaqus on a shared folder, for example, /share/apps, check with your system administrator for the shared folder on your system. All the abaqus files will be accessable from all nodes if you choose this way.