Re: [slurm-users] Using GRES to manage GPUs, but unable to assign specific CPUs to specific GPUs

2018-09-17 Thread Julie Bernauer
Hi Randy, This is expected on an HT machine, like on the one described below. If you run lstopo, you see: L2 L#5 (256KB) + L1d L#5 (32KB) + L1i L#5 (32KB) + Core L#5 PU L#10 (P#5) PU L#11 (P#45) Slurm uses the logical cores so 10 and 11 gives you "physical" cores 5 and 45.

Re: [slurm-users] Slurm 18.08 X11 Errors

2018-09-17 Thread Nathan R.M. Crawford
Hi Luca, I hit something similar with 17.11.9-2 (and earlier 17.11 releases) on CentOS 7.5, but it may not be your issue. For us, the xauth entry generated through "srun --x11" only has the short hostname, but the entry created with "ssh -X" has the full host name. Example lines from "xauth list

[slurm-users] Anyone see odd job array dependency issues?

2018-09-17 Thread Eli V
I'm seeing a weird issue(originally with 17.02 and still after upgrading to 18.08) where occasionally job arrays created with -d aftercorr seem to be getting mixed up in the slurm controller and the wrong jobs are getting started and cancelled. Just created a bug for it: https://bugs.schedmd.com/sh

Re: [slurm-users] Slurm 18.08 X11 Errors

2018-09-17 Thread Luca Cenzato
Hi Loris, yes, I am using "-X"; it's aliased and I forgot to wrote "\ssh -X" in the tests. I forgot also that: all servers are running CentOs 7.5; I've updated from 17.11.9-2 (not tested, only checked the installation procedure) Thanks! Luca Il giorno lun 17 set 2018 alle ore 07:51 Loris Benne