Merge pull request #129 from stackhpc/fix/slurmdbd-restart

Add wait_for slurmdbd port

Author: sjpb

handlers/main.yml

@@ -7,7 +7,7 @@
   when: openhpc_slurm_service_started | bool
 
 # NOTE: we need this running before slurmctld start
-- name: Restart slurmdbd service
+- name: Issue slurmdbd restart command
   service:
     name: "slurmdbd"
     state: restarted
@@ -17,6 +17,18 @@
     - openhpc_slurm_service_started | bool
     - openhpc_slurmdbd_host in play_hosts
   run_once: true
+  listen: Restart slurmdbd service
+
+- name: Check slurmdbd actually restarted
+  wait_for:
+    port: "{{ openhpc_slurmdbd_port }}"
+  delegate_to: "{{ openhpc_slurmdbd_host }}"
+  run_once: true
+  when:
+    - openhpc_slurmdbd_host is defined
+    - openhpc_slurm_service_started | bool
+    - openhpc_slurmdbd_host in play_hosts
+  listen: Restart slurmdbd service
 
 # NOTE: we need this running before slurmd
 # Allows you to reconfigure slurmctld from another host

tasks/runtime.yml

@@ -163,16 +163,16 @@
     enabled: "{{ openhpc_slurm_service_enabled | bool }}"
     state: "{{ 'started' if openhpc_slurm_service_started | bool else 'stopped' }}"
 
+- name: flush handler
+  meta: flush_handlers # as then subsequent "ensure" is a no-op if slurm services bounced
+
 - name: Ensure slurmdbd is started and running
   service:
     name: slurmdbd
     enabled: "{{ openhpc_slurm_service_enabled | bool }}"
     state: "{{ 'started' if openhpc_slurm_service_started | bool else 'stopped' }}"
   when: openhpc_enable.database | default(false) | bool
 
-- name: flush handler
-  meta: flush_handlers # as then subsequent "ensure" is a no-op if slurm services bounced
-
 - name: Ensure Slurm service state
   service:
     name: "{{ openhpc_slurm_service }}"