Re: [slurm-users] Getting nodes in a partition
Hi Mahmood,
I needed something similar and here is what I was suggested to do. You need
to further modify the node list to be bash-friendly using curly-bracketed
ranges:
MYPART=debug
NODES=`scontrol show partition $MYPART | grep -w Nodes | cut -d '=' -f 2 |
sed -r -e 's:[[](.*)[]]:{\1}:' -e 's:([0-9]+)[-]([0-9]+):{\1..\2}:g' | tr
-s {}`
echo $NODES
cluster-{105..120}
eval echo $NODES
cluster-105 cluster-106 cluster-107 cluster-108 cluster-109 cluster-110
cluster-111 cluster-112 cluster-113 cluster-114 cluster-115 cluster-116
cluster-117 cluster-118 cluster-119 cluster-120
It doesnt work if you have comma separated node lists, e.g.
cluster-[105-121],cluster-[125-140], but it does work for comma separated
ranges, e.g. cluster-[109,115-116,118-121,130,138]. In the former case, you
can split the node lists into two or more, and do the conversion for each
sub-list.
Best regards,
Renat.
On 18 May 2018 at 09:11, Mahmood Naderan wrote:
> Hi,
> Is there any slurm variable to read the node names of a partition?
> There is an MPI option --hostfile which we can write the node names.
> I want to use something like this in the sbatch script:
>
>
> #SBATCH --partition=MYPART
> ... --hostfile $SLURM_NODES_IN_PARTITION
>
> I can manually manipulate the output of scontrol to extract node
> names. Like this:
>
> [mahmood@rocks7 ~]$ scontrol show partition MYPART | grep -w Nodes |
> cut -d '=' -f 2
> compute-0-[4-6]
>
> But that basically is not
> compute-0-4
> compute-0-5
> compute-0-6
>
> Which I have to post process more. Any better idea?
>
>
>
> Regards,
> Mahmood
>
>
Re: [slurm-users] Getting nodes in a partition
Hi Mahmood, slurm comes with hostlist extraction: [mahmood@rocks7 ~]$ scontrol show hostnames $(scontrol show partition MYPART | grep -w Nodes | cut -d '=' -f 2) This would then be compute-0-4 compute-0-5 compute-0-6 Best Marcus On 05/18/2018 09:11 AM, Mahmood Naderan wrote: Hi, Is there any slurm variable to read the node names of a partition? There is an MPI option --hostfile which we can write the node names. I want to use something like this in the sbatch script: #SBATCH --partition=MYPART ... --hostfile $SLURM_NODES_IN_PARTITION I can manually manipulate the output of scontrol to extract node names. Like this: [mahmood@rocks7 ~]$ scontrol show partition MYPART | grep -w Nodes | cut -d '=' -f 2 compute-0-[4-6] But that basically is not compute-0-4 compute-0-5 compute-0-6 Which I have to post process more. Any better idea? Regards, Mahmood -- Marcus Wagner, Dipl.-Inf. IT Center Abteilung: Systeme und Betrieb RWTH Aachen University Seffenter Weg 23 52074 Aachen Tel: +49 241 80-24383 Fax: +49 241 80-624383 wag...@itc.rwth-aachen.de www.itc.rwth-aachen.de
Re: [slurm-users] Getting nodes in a partition
Does sinfo -h -O nodehost -p partition | sort help? Also scontrol show hostname nodelist where nodelist is compute-0-\[4-6\] would work. Regards Henk -Original Message- From: slurm-users On Behalf Of Mahmood Naderan Sent: 18 May 2018 08:12 To: Slurm User Community List Subject: [slurm-users] Getting nodes in a partition Hi, Is there any slurm variable to read the node names of a partition? There is an MPI option --hostfile which we can write the node names. I want to use something like this in the sbatch script: #SBATCH --partition=MYPART ... --hostfile $SLURM_NODES_IN_PARTITION I can manually manipulate the output of scontrol to extract node names. Like this: [mahmood@rocks7 ~]$ scontrol show partition MYPART | grep -w Nodes | cut -d '=' -f 2 compute-0-[4-6] But that basically is not compute-0-4 compute-0-5 compute-0-6 Which I have to post process more. Any better idea? Regards, Mahmood
Re: [slurm-users] Getting nodes in a partition
Hi,
It seems to me that the sinfo command is the simplest solution to
listing hosts in a partition. Here is an example:
# sinfo -N -p xeon8_48
NODELIST NODES PARTITION STATE
d001 1 xeon8_48 idle
d002 1 xeon8_48 idle
d003 1 xeon8_48 idle
d004 1 xeon8_48 idle
d005 1 xeon8_48 idle
d006 1 xeon8_48 idle
d007 1 xeon8_48 idle
d008 1 xeon8_48 idle
d009 1 xeon8_48 idle
d010 1 xeon8_48 idle
d011 1 xeon8_48 idle
d012 1 xeon8_48 idle
d013 1 xeon8_48 idle
d014 1 xeon8_48 idle
d015 1 xeon8_48 idle
d016 1 xeon8_48 idle
d017 1 xeon8_48 idle
d018 1 xeon8_48 idle
d019 1 xeon8_48 idle
d020 1 xeon8_48 idle
d021 1 xeon8_48 idle
d022 1 xeon8_48 idle
The output of sinfo can be formatted in many ways, please see the man-page.
I have some remarks on host lists in my Wiki page
https://wiki.fysik.dtu.dk/niflheim/SLURM#expanding-and-collapsing-host-lists
/Ole
On 05/18/2018 10:12 AM, Renat Yakupov wrote:
Hi Mahmood,
I needed something similar and here is what I was suggested to do. You
need to further modify the node list to be bash-friendly using
curly-bracketed ranges:
MYPART=debug
NODES=`scontrol show partition $MYPART | grep -w Nodes | cut -d '=' -f 2
| sed -r -e 's:[[](.*)[]]:{\1}:' -e 's:([0-9]+)[-]([0-9]+):{\1..\2}:g' |
tr -s {}`
echo $NODES
cluster-{105..120}
eval echo $NODES
cluster-105 cluster-106 cluster-107 cluster-108 cluster-109
cluster-110 cluster-111 cluster-112 cluster-113 cluster-114 cluster-115
cluster-116 cluster-117 cluster-118 cluster-119 cluster-120
It doesnt work if you have comma separated node lists, e.g.
cluster-[105-121],cluster-[125-140], but it does work for comma
separated ranges, e.g. cluster-[109,115-116,118-121,130,138]. In the
former case, you can split the node lists into two or more, and do the
conversion for each sub-list.
Best regards,
Renat.
On 18 May 2018 at 09:11, Mahmood Naderan > wrote:
Hi,>
Regards,
Mahmood
--
Ole Holm Nielsen
PhD, Senior HPC Officer
Department of Physics, Technical University of Denmark,
Building 307, DK-2800 Kongens Lyngby, Denmark
E-mail: ole.h.niel...@fysik.dtu.dk
Homepage: http://dcwww.fysik.dtu.dk/~ohnielse/
Tel: (+45) 4525 3187 / Mobile (+45) 5180 1620
Is there any slurm variable to read the node names of a partition?
There is an MPI option --hostfile which we can write the node names.
I want to use something like this in the sbatch script:
#SBATCH --partition=MYPART
... --hostfile $SLURM_NODES_IN_PARTITION
I can manually manipulate the output of scontrol to extract node
names. Like this:
[mahmood@rocks7 ~]$ scontrol show partition MYPART | grep -w Nodes |
cut -d '=' -f 2
compute-0-[4-6]
But that basically is not
compute-0-4
compute-0-5
compute-0-6
Which I have to post process more. Any better idea?
Re: [slurm-users] Getting nodes in a partition
>slurm comes with hostlist extraction: > >[mahmood@rocks7 ~]$ scontrol show hostnames $(scontrol show partition MYPART | >grep -w Nodes | cut -d '=' -f 2) Excellent. That works. Many thanks... I wrote the following method before that! [mahmood@rocks7 ~]$ cat nodes.sh #!/bin/bash NAMES=`scontrol show partition NOLIMIT | grep -w Nodes | cut -d '=' -f 2` echo $NAMES BASENAME=`echo $NAMES | cut -d '[' -f 1` RANGE=`echo $NAMES | cut -d '[' -f2 | cut -d ']' -f1` START=`echo $RANGE | cut -d '-' -f1` END=`echo $RANGE | cut -d '-' -f2` for (( i=$START; i <= $END; i++ )); do echo "$BASENAME$i" done [mahmood@rocks7 ~]$ ./nodes.sh compute-0-[4-6] compute-0-4 compute-0-5 compute-0-6 Regards, Mahmood
Re: [slurm-users] Getting nodes in a partition
>Does > >sinfo -h -O nodehost -p partition | sort > >help? Yes that also works. Thanks Henk. Regards, Mahmood
Re: [slurm-users] Getting nodes in a partition
I saw you got some good answers, but a quick note on mpi. For some of them, you are compiling it yourself, they can be "slurm-aware" (eg: openmpi). Then when you do 'mpirun' it automatically knows your inherited hostlist and you need do nothing extra when running. Brian Andrus On 5/18/2018 12:11 AM, Mahmood Naderan wrote: Hi, Is there any slurm variable to read the node names of a partition? There is an MPI option --hostfile which we can write the node names. I want to use something like this in the sbatch script: #SBATCH --partition=MYPART ... --hostfile $SLURM_NODES_IN_PARTITION I can manually manipulate the output of scontrol to extract node names. Like this: [mahmood@rocks7 ~]$ scontrol show partition MYPART | grep -w Nodes | cut -d '=' -f 2 compute-0-[4-6] But that basically is not compute-0-4 compute-0-5 compute-0-6 Which I have to post process more. Any better idea? Regards, Mahmood
