Further nodes available in the PBSPro experimental environment

Switching Torque @arien to PBS Pro @arien-pro is scheduled for next week.

Almost all resources were moved to the PBS Pro and Torque's queues have been disabled yesterday afternoon.

Most of frontends were set to PBS Pro environment @arien-pro, all the others will be switched probabely on Monday next week.

Actual information: https://wiki.metacentrum.cz/wiki/Frontend

Please, do not use the old Torque environment @arien for new jobs, send them to directly to PBS Pro @arien-pro.

If you are using a frontend without default PBS Pro setting, it is necessary to activate the PBS Pro environment on the frontend by the command:

   module add pbspro-client

In CERIT-SC, there are available only a few special machines in the PBS Pro environment (@wagap-pro) -- uv2 (unga a urgu) and XEON Phi (phi) now. Other machines will be switched to PBS Pro a few months later.

Please note:

PBSPro server: arien-pro.ics.muni.cz (server is not directly accessible)
GPU cluster is accessible via dedicated gpu@arien-pro.ics.muni.cz and gpu_long@arien-pro.ics.muni.cz queues
PBS Pro comes with new qsub syntax "select" which differs in major aspects from old Torque syntax -- request for job with two processors on each of 3 chunks (nodes), 1 GB of RAM, 1 GB of local scratch, 1 hour walltime, 1 licence for Matlab:
- ```
  qsub -l select=3:ncpus=2:mem=1gb:scratch_local=1gb -l walltime=1:00:00 -l matlab=1 skript.sh
```
Scratch requirement must be stated individually. When requesting scratch, it is necessary to specify its type. There is no default type of scratch! (scratch_local=1gb)!!!
Qsub assembler for PBSPro and other information from PBSPro were integrated to PBSMon: https://metavo.metacentrum.cz/en/state/personal
Documentation for new scheduling system PBS Professional can be found at wiki https://wiki.metacentrum.cz/wiki/PBS_Professional

With best regards,

MetaCentrum
MetaCentrum

Ivana Křenková, Sat Mar 25 21:39:00 CET 2017