Run zenith-client init with a command to get foreground behaviour

Author: Matt Pryor

group_vars/openstack.yml

@@ -47,5 +47,3 @@ cluster_groups_validation:
 cluster_groups_zenith:
   # Any hosts in the grafana group should go in the zenith group
   zenith: [grafana]
-  # Any hosts in the zenith group should have podman installed
-  podman: [zenith]

roles/zenith_proxy/tasks/main.yml

@@ -44,15 +44,16 @@
   become_user: "{{ zenith_proxy_podman_user }}"
 
 - name: Initialise Zenith client
-  containers.podman.podman_container:
-    name: "{{ zenith_proxy_service_name }}-init"
-    image: "{{ zenith_proxy_client_image }}"
-    command: [zenith-client, init]
-    detach: no
-    tty: yes
-    volumes:
-      - /etc/zenith/{{ zenith_proxy_service_name }}:/etc/zenith:ro
-      - "{{ zenith_proxy_service_name }}-ssh:/home/zenith/.ssh"
+  # Use a foreground command rather than the podman_container module as I could not
+  # work out the combination of parameters that produced the desired behaviour :-(
+  command: >-
+    podman run
+      --name {{ zenith_proxy_service_name }}-init
+      --replace
+      --volume /etc/zenith/{{ zenith_proxy_service_name }}:/etc/zenith:ro
+      --volume {{ zenith_proxy_service_name }}-ssh:/home/zenith/.ssh
+      {{ zenith_proxy_client_image }}
+      zenith-client init
   become: true
   become_user: "{{ zenith_proxy_podman_user }}"
   register: zenith_proxy_client_init

slurm-infra.yml

@@ -58,6 +58,13 @@
         groups: "{{ hostvars[item].group_names | terraform_infra_expand_groups(cluster_groups_validation) }}"
       loop: "{{ groups.get('cluster', []) }}"
       when: cluster_run_validation | default(false) | bool
+    # If Zenith is enabled, add the zenith groups to hosts
+    - name: Add clusters hosts to Zenith groups
+      add_host:
+        name: "{{ item }}"
+        groups: "{{ hostvars[item].group_names | terraform_infra_expand_groups(cluster_groups_zenith) }}"
+      loop: "{{ groups.get('cluster', []) }}"
+      when: zenith_subdomain_monitoring is defined
 
 # Ensure that the hosts in the cluster can all refer to each other by their hostname
 - hosts: cluster
@@ -115,16 +122,18 @@
 
 # Deploy the monitoring stack first
 - import_playbook: vendor/stackhpc/ansible-slurm-appliance/ansible/monitoring.yml
-# Ensure that podman is installed on the host running grafana
-- hosts: grafana,!podman
+
+# Configure the Zenith clients that are required
+# First, ensure that podman is installed on all hosts that will run Zenith clients
+- hosts: zenith,!podman
   tasks:
     - import_role:
         name: podman
         tasks_from: prereqs.yml
     - import_role:
         name: podman
         tasks_from: config.yml
-# Then deploy the Zenith client onto the Grafana hosts to talk to it
+# Deploy the Zenith client for Grafana
 - hosts: grafana
   tasks:
     - include_role:

ui-meta/slurm-infra.yml

@@ -72,7 +72,7 @@ usage_template: |-
   Access to the cluster is via the external IP using SSH as the `rocky` user:
 
   ```
-  $ ssh rocky@{{ cluster.outputs.cluster_access_ip | default('<cluster ip>') }}
+  $ ssh rocky@{{ cluster.outputs.cluster_access_ip | default('[cluster ip]') }}
   [rocky@{{ cluster.name }}-login-0 ~]$ sinfo
   PARTITION AVAIL  TIMELIMIT  NODES  STATE NODELIST
   compute*     up 60-00:00:0    {{ "%3s" | format(cluster.parameter_values.compute_count) }}   idle {{ cluster.name }}-compute-[0-{{ cluster.parameter_values.compute_count - 1 }}]