# Ansible

https://github.com/RyaxTech/ansible-hpc-cluster

~~~
---
- name: Deploy Slurm Cluster
hosts: all
become: yes
vars:
control_node: hpc1
compute_nodes: hpc2
slurm_version: 23.11.4
munge_key_path: /etc/munge/munge.key

tasks:
- name: Install common dependencies
apt:
name: "{{ item }}"
state: present
loop:
- wget
- build-essential
- libmunge-dev
- nfs-common

- name: Install Munge on all nodes
apt:
name: munge
state: present

- name: Generate munge key on control node
command: /usr/sbin/create-munge-key -f
when: inventory_hostname == control_node
notify: Restart Munge

- name: Copy munge key to compute nodes
ansible.builtin.copy:
src: "{{ munge_key_path }}"
dest: "{{ munge_key_path }}"
owner: munge
group: munge
mode: '0400'
when: inventory_hostname != control_node

- name: Configure Control Node
hosts: hpc1
tasks:
- name: Install Slurm Controller
apt:
name: slurmctld
state: present

- name: Create Slurm config
template:
src: slurm.conf.j2
dest: /etc/slurm/slurm.conf
notify: Restart Slurm Controller

- name: Start NFS server
service:
name: nfs-kernel-server
state: started
enabled: yes

- name: Configure Compute Nodes
hosts: all
tasks:
- name: Mount NFS share
mount:
path: /etc/slurm
src: "{{ control_node }}:/etc/slurm"
fstype: nfs
state: mounted

- name: Install Slurm Compute Daemon
apt:
name: slurmd
state: present

- name: Start Slurmd
service:
name: slurmd
state: restarted
enabled: yes

handlers:
- name: Restart Munge
service:
name: munge
state: restarted

- name: Restart Slurm Controller
service:
name: slurmctld
state: restarted

~~~