In your slurm.conf file do you intentionally put # infront of NodeAddr. Can you
remove it and give it another try?
Said.
________________________________
From: 허웅 <hoewoongg...@naver.com>
Sent: Friday, July 28, 2017 5:37:35 PM
To: slurm-dev
Subject: [slurm-dev] very strange situation in using slurm
I'm very new to using slurm.
and My cluster look like :
ControlNode : GO1
ComputeNodes : GO[1-5]
I want to allocate some job to specific node.
So i tried this command :
$ srun -w go2 hostname
I expected that will print "GO2"
but printed "GO1"
really strange thing is that I excute this command after what i did(srun -w go2
hostname):
$ sinfo
PARTITION AVAIL TIMELIMIT NODES STATE NODELIST
debug* up infinite 1 alloc go2
debug* up infinite 4 idle go[1,3-5]
$ squeue
JOBID PARTITION NAME USER ST TIME NODES
NODELIST(REASON)
351 debug sleep root R 0:03 1 go2
I think the controller have requested to specific node(in this case, go2) But,
the result is "GO1"
what did i wrong?
[https://mail.naver.com/readReceipt/notify/?img=3eRTWX%2B0Wr2spzeTpr%2B5hAtrMxtwKqF4aA2XKzkvKxEdaAkSKo3oFztqKoK%2FMqCgMX%2B0Moud74lR74lcWNFlbX30WLloWrdQaXFn%2BBiR7630%2BvIqMrC0pz9v74FTWt%3D%3D.gif]
