2

I have a user account on a very big cluster. I have previous experience with Grid Engine and want to use the cluster for array jobs.

The documentation tells me to use "qsub" for load balancing / submission of many jobs. Therefore I assumed this means the cluster has Grid Engine.

However all my Grid Engine scripts failed to run. I checked the documentation and it is a bit weird. Now I slowly suspect that this cluster does not actually have Grid Engine, maybe it's running something called Torque (?!). The whole terminology in the man pages is a bit weird for me as a Grid Engine user, for example they talk about "bulk jobs" instead of "array jobs". There is no referral to variables on which I rely on, like SGE_TASK_ID etc. Instead they refer to variables starting with PBS_. Still, there are qsub and qstat commands.

Also qsub behaves differently, apparently it is not possible to specifiy the command line parameters with bash-script comments etc.

There is a documentation for the cluster system, but it does not say what the DRM middleware actually is - it refers to the entire DRM system simply as "qsub".

I tried

qsub --version
qsub: 1.2 2010/8/17

I am not sure what I am actually running when I invoke qsub on that cluster!

My question is, how can I find out if I am running Grid Engine or Torque (or whatever it is), and which version?

1

Although this does not answer your question directly, I'd like to point out that qsub/qstat/qselect... commands are all part of interface defined in POSIX Batch Environment Services specification, hence the PBS_ prefix for variables. Restricting your BE interface to the standartized subset is supposed to make your scripts independent of a specific grid implementation.

Hope this helps.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy