Merge upstream and cherry-pick fixes (#3)

* Added cluster creation to slurmdb install

* Refactored order in which cluster is created

* review order of restarting service

should follow this order when restarting :
- slurmdbd
- slurmctld
- slurmd

* Fix syntax error in handlers

* Fix typo in handler name

(cherry picked from commit 712cf32)

* Remove key 'SlurmctldPidFile' from __slurmdbd_config_default

(cherry picked from commit 936e64a)

---------

Co-authored-by: slugger70 <[email protected]>
Co-authored-by: Christian IUGA <[email protected]>
Co-authored-by: Nate Coraor <[email protected]>
Co-authored-by: Nuwan Goonasekera <[email protected]>
Co-authored-by: Christopher Lilienthal <[email protected]>
Co-authored-by: cat-bro <[email protected]>

defaults/main.yml

@@ -17,6 +17,10 @@ slurmd_service_name: slurmd
 slurmctld_service_name: slurmctld
 slurmdbd_service_name: slurmdbd
 
+#Cluster name for slurm config. This is required to correctly setup slurmdbd and attune it to the slurm config.
+__slurm_cluster_name: cluster
+__cluster_not_setup: true #Default value. Is modified if cluster already exists.
+
 slurm_start_services: true
 
 
@@ -48,7 +52,7 @@ __slurm_config_default:
   AuthType: auth/munge
   CryptoType: crypto/munge
   SlurmUser: "{{ __slurm_user_name }}"
-  ClusterName: cluster
+  ClusterName: "{{ __slurm_cluster_name }}"
   # default is proctrack/cgroup which is the best but also less than 100% chance of working e.g. in docker
   ProctrackType: proctrack/pgid
   # slurmctld options
@@ -93,6 +97,5 @@ __slurmdbd_config_default:
   AuthType: auth/munge
   DbdPort: 6819
   SlurmUser: "{{ __slurm_user_name }}"
-  SlurmctldPidFile: "{{ __slurm_run_dir ~ '/slurmdbd.pid' if __slurm_debian else omit }}"
   LogFile: "{{ __slurm_log_dir ~ '/slurmdbd.log' if __slurm_debian else omit }}"
 __slurmdbd_config_merged: "{{ __slurmdbd_config_default | combine(slurmdbd_config | default({})) }}"

handlers/main.yml

@@ -4,20 +4,20 @@
     name: munge
     state: restarted
 
-- name: Reload slurmd
+- name: Reload slurmdbd
   ansible.builtin.service:
-    name: "{{ slurmd_service_name }}"
+    name: "{{ slurmdbd_service_name }}"
     state: reloaded
-  when: "slurm_start_services and ('slurmexechosts' in group_names or 'exec' in slurm_roles)"
+  when: "slurm_start_services and ('slurmdbdservers' in group_names or 'dbd' in slurm_roles)"
 
-- name: Restart slurmd
+- name: Restart slurmdbd
   ansible.builtin.systemd:
-    name: "{{ slurmd_service_name }}"
+    name: "{{ slurmdbd_service_name }}"
     state: restarted
     masked: no
     enabled: yes
     daemon_reload: yes
-  when: "slurm_start_services and ('slurmexechosts' in group_names or 'exec' in slurm_roles)"
+  when: "slurm_start_services and ('slurmservers' in group_names or 'controller' in slurm_roles)"
 
 - name: Reload slurmctld
   ansible.builtin.service:
@@ -34,17 +34,14 @@
     daemon_reload: yes
   when: "slurm_start_services and ('slurmservers' in group_names or 'controller' in slurm_roles)"
 
-- name: Restart slurmdbd
-  ansible.builtin.systemd:
-    name: "{{ slurmdbd_service_name }}"
-    state: restarted
-    masked: no
-    enabled: yes
-    daemon_reload: yes
-  when: "slurm_start_services and ('slurmservers' in group_names or 'controller' in slurm_roles)"
-
-- name: Reload slurmdbd
+- name: Reload slurmd
   ansible.builtin.service:
-    name: "{{ slurmdbd_service_name }}"
+    name: "{{ slurmd_service_name }}"
     state: reloaded
-  when: "slurm_start_services and ('slurmdbdservers' in group_names or 'dbd' in slurm_roles)"
+  when: "slurm_start_services and ('slurmexechosts' in group_names or 'exec' in slurm_roles)"
+
+- name: Restart slurmd
+  ansible.builtin.service:
+    name: "{{ slurmd_service_name }}"
+    state: restarted
+  when: "slurm_start_services and ('slurmexechosts' in group_names or 'exec' in slurm_roles)"

tasks/main.yml

@@ -43,3 +43,7 @@
     enabled: true
     state: started
   when: "slurm_start_services and ('slurmexechosts' in group_names or 'exec' in slurm_roles)"
+
+- name: Setup cluster on slurmdb
+  include_tasks: slurmdbd_cluster.yml
+  when: "slurm_start_services and ('slurmdbdservers' in group_names or 'dbd' in slurm_roles)"

tasks/slurmdbd_cluster.yml

@@ -0,0 +1,20 @@
+---
+
+- name: Check for existence of cluster in db.
+  register: cluster_check
+  shell: "sacctmgr -n list cluster | cut -f 4 -d ' '"
+  become: yes
+  become_user: root
+
+- name: set cluster_check_boolean
+  set_fact:
+      __cluster_not_setup: false
+  when: cluster_check.stdout == "cluster"
+
+- name: Create the slurmdbd cluster
+  command: sacctmgr -i -n add cluster {{ __slurm_cluster_name }}
+  become: yes
+  become_user: root
+  notify:
+    - Reload slurmdbd
+  when: __cluster_not_setup