Open xingwu2 opened 7 years ago
Hi,
Sorry for not replying sooner. There isn't any support for Torque by default. There was some work done in getting jobTree (our scheduling abstraction) to support Torque, but I wasn't involved in that and can't say whether it actually worked. I think it been left unfinished, or had bugs, because it never was merged into jobTree.
I notice there is a "torque" branch of jobTree -- https://github.com/benedictpaten/jobTree/tree/torque -- maybe changing to the submodules/jobTree directory and checking out the torque branch will provide (experimental) support? I wouldn't hold out much hope of that working, though.
Torque never panned out because it didn't handle the loads of tiny jobs JobTree spawns well enough. It would take hours just to run small examples. We talked about adding some kind of job batching support to get around this but I don't think it got too far.
On Tue, Jan 3, 2017 at 7:56 PM, Joel Armstrong notifications@github.com wrote:
Hi,
Sorry for not replying sooner. There isn't any support for Torque by default. There was some work done in getting jobTree (our scheduling abstraction) to support Torque, but I wasn't involved in that and can't say whether it actually worked. I think it been left unfinished, or had bugs, because it never was merged into jobTree.
I notice there is a "torque" branch of jobTree -- https://github.com/ benedictpaten/jobTree/tree/torque -- maybe changing to the submodules/jobTree directory and checking out the torque branch will provide (experimental) support? I wouldn't hold out much hope of that working, though.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/glennhickey/progressiveCactus/issues/71#issuecomment-270268713, or mute the thread https://github.com/notifications/unsubscribe-auth/AA2_7muPwkMEqlXxGNE2A0Fld8YJ3YqBks5rOu5KgaJpZM4LTRDL .
I've been working on a SLURM option for jobTree based on the torque branch.
You can have a look here: https://github.com/ljyanesm/jobTree
It's WIP also, we need to specify the queue on our systems and there is no configuration for that atm so it's hardcoded.
Hi @ljyanesm , I'm interested in using the progressiveCactus on SLURM, I was wondering if you could provide a guidance, please.
Thanks in advance. I look forward to hearing from you.
Hi,
Have a look at the following commit: https://github.com/ljyanesm/jobTree/commit/6aa62436b59ad3759110286995644b613da484c1
There you can find the modifications we had to do in order to use SLURM at EI. You still need to be aware that the configuration files have to express the memory requirement values in MB instead of bytes as the original ones do.
The other bit you should be mindful of is the partition (-p) if your site doesn't have a long enough "default" partition you can set it when submitting the jobs as is done on batchSystems/drmaaSlurm.py
Hi,
I am using a PBS/Torque cluster. I wonder do you have a modified release that supports the PBS/Torque or Slurm cluster?
Best
Xing Wu