Slurm is a job scheduler for high-performance computing clusters
313
Slurm is a job scheduler for high-performance computing clusters. It is designed to manage large clusters of computers and provides a simple interface for users to submit jobs to the cluster.
JWT_KEY:
label: "JWT Key"
hint: "More than 32 characters"
default: "auto"
required: true
MASTER_HOSTNAME:
label: "Master Hostname"
hint: "IP address or hostname"
default: "localhost"
required: true
WORKER_HOSTNAMES:
label: "Worker Hostnames"
hint: "Comma-separated list"
default: "localhost"
required: true
PARTITION_NAMES:
label: "Partition Names"
hint: "Comma-separated list"
default: "main"
required: true
VOLUME:
label: "Volume"
hint: "Starts with /volume"
default: "/volume"
required: true
---
name: slurm
services:
instance:
image: diphyx/slurm
privileged: true
environment:
- JWT_KEY={{ JWT_KEY }}
- MASTER_HOSTNAME={{ MASTER_HOSTNAME }}
- WORKER_HOSTNAMES={{ WORKER_HOSTNAMES }}
- PARTITION_NAMES={{ PARTITION_NAMES }}
- DXF_PROXY_MAIN_PORT=6820
volumes:
- {{ VOLUME }}:/volume
docker run -it --rm --name slurm -p 6817:6817 -p 6818:6818 -p 6819:6819 -p 6820:6820 --privileged diphyx/slurm:latest MASTER,WORKER
docker run -it --rm --name slurm-master -p 6817:6817 -p 6819:6819 -p 6820:6820 --privileged diphyx/slurm:latest MASTER
docker run -it --rm --name slurm-worker -p 6819:6819 --privileged diphyx/slurm:latest WORKER
mkdir -p /volume/jobs
{
echo '#!/bin/bash'
echo '#SBATCH --partition=main'
echo '#SBATCH --job-name=demo'
echo '#SBATCH --output=output.log'
echo 'echo "Sleep to 5 seconds"'
echo 'sleep 5'
echo 'echo "Done"'
} >> /volume/jobs/demo.sh
sbatch /volume/jobs/demo.sh
This image only works on the DiPhyx platform.
Content type
Image
Digest
sha256:f0d0f9e41…
Size
168.8 MB
Last updated
over 1 year ago
docker pull diphyx/slurm