Prerequisites

• kubectl version v1.11.3+.
• buildah version v1.33.10+
• Access to a Kubernetes v1.11.3+ cluster.

Features

Slurm on Kubernetes provides the following features:

• Resource Management: Efficiently manages resources in a Kubernetes cluster, ensuring optimal utilization.
• Job Scheduling: Advanced job scheduling capabilities to handle various types of workloads.
• Scalability: Easily scales to accommodate growing workloads and resources.
• High Availability: Supports high availability configurations to ensure continuous operation.
• Multi-User Support: Allows multiple users to submit and manage their jobs concurrently.
• Integration with MPI Libraries: Supports both Open MPI and Intel MPI libraries for parallel computing.
• Customizable: Using values.yaml file, you can customizable a slurm cluster, fitting specific needs and configurations.
• Separated munged daemon
• Support GPU nodes deployment
• Running on Cgroup v1/v2

Usage

if you wanna change slurm configuration ,please check slurm configuration generator, check link

• for github helm user
  1. get helm repo and update

     helm repo add ay-helm-mirror https://aaronyang0628.github.io/helm-chart-mirror/charts
     

  2. install slurm chart

     helm install slurm ay-helm-mirror/chart -f charts/values.yaml --version 1.0.10
     

• for artifact helm user
  1. get helm repo and update

     helm repo add ay-helm-mirror https://aaronyang0628.github.io/helm-chart-mirror/charts
     

  2. install slurm chart

     helm install slurm ay-helm-mirror/chart -f charts/values.yaml --version 1.0.10

  Or you can get template values.yaml from link
• for opertaor user
  1. test pull an image and apply

     podman pull ghcr.io/aaronyang0628/slurm-operator:25.05
     

  2. deploy slurm operator

     kubectl apply -f https://raw.githubusercontent.com/AaronYang0628/slurm-on-k8s/refs/heads/main/operator/dist/install.yaml
     

  3. apply CRD slurmdeployment

     kubectl apply -f https://raw.githubusercontent.com/AaronYang0628/helm-chart-mirror/refs/heads/main/templates/slurm/slurmdeployment.zj.values.yaml
     

Manage Your Slurm Cluster

• check cluster status

  kubectl get slurmdep slurmdeployment-sample
  kubectl -n slurm get pods -w

When everything is ready, you can login your cluster and submit jobs.

• Add PubKeys to login node

  you can edit `auth.ssh.configmap.perfabPubKeys` in the file chart/values.yaml, adding your public keys to the end 
  Or you can edit `spec.values.auth.ssh.configmap.perfabPubKeys` in your slurmdeployment CRD

• reapply your chart or CRD

• login your cluster

  kubectl -n slurm exec -it deploy/slurm-login -c login -- bin/bash

  Or

  ssh root@slurm-login.svc.cluster.local