Ported monitoring stack to k3s

ansible/.gitignore

@@ -70,6 +70,8 @@ roles/*
 !roles/k3s/**
 !roles/k9s/
 !roles/k9s/**
+!roles/kube_prometheus_stack
+!roles/kube_prometheus_stack/**
 !roles/lustre/
 !roles/lustre/**
 !roles/dnf_repos/

ansible/fatimage.yml

@@ -150,12 +150,6 @@
         tasks_from: install.yml
       when: "'filebeat' in group_names"
 
-    - import_role:
-      # can't only run cloudalchemy.node_exporter/tasks/install.yml as needs vars from preflight.yml and triggers service start
-      # however starting node exporter is ok
-        name: cloudalchemy.node_exporter
-      when: "'node_exporter' in group_names"
-
     - name: openondemand exporter
       dnf:
         name: ondemand_exporter
@@ -168,72 +162,16 @@
       vars:
         slurm_exporter_state: stopped
       when: "'slurm_exporter' in group_names"
-
+    
 - hosts: prometheus
   become: yes
   gather_facts: yes
   tasks:
-    - import_role:
-        name: cloudalchemy.prometheus
-        tasks_from: preflight.yml
-
-    # can't run cloudalchemy.prometheus/tasks/install.yml as it triggers a unit start
-    # so below is a partial extraction of this:
-    - name: create prometheus system group
-      group:
-        name: prometheus
-        system: true
-        state: present
-
-    - name: create prometheus system user
-      user:
-        name: prometheus
-        system: true
-        shell: "/usr/sbin/nologin"
-        group: prometheus
-        createhome: false
-        home: "{{ prometheus_db_dir }}"
-
-    - name: download prometheus binary to local folder
-      become: false
-      get_url:
-        url: "https://github.com/prometheus/prometheus/releases/download/v{{ prometheus_version }}/prometheus-{{ prometheus_version }}.linux-{{ go_arch }}.tar.gz"
-        dest: "/tmp/prometheus-{{ prometheus_version }}.linux-{{ go_arch }}.tar.gz"
-        checksum: "sha256:{{ __prometheus_checksum }}"
-      register: _download_archive
-      until: _download_archive is succeeded
-      retries: 5
-      delay: 2
-
-    - name: unpack prometheus binaries
-      become: false
-      unarchive:
-        remote_src: yes
-        src: "/tmp/prometheus-{{ prometheus_version }}.linux-{{ go_arch }}.tar.gz"
-        dest: "/tmp"
-        creates: "/tmp/prometheus-{{ prometheus_version }}.linux-{{ go_arch }}/prometheus"
-
-    - name: propagate official prometheus and promtool binaries
-      copy:
-        remote_src: yes
-        src: "/tmp/prometheus-{{ prometheus_version }}.linux-{{ go_arch }}/{{ item }}"
-        dest: "{{ _prometheus_binary_install_dir }}/{{ item }}"
-        mode: 0755
-        owner: root
-        group: root
-      with_items:
-        - prometheus
-        - promtool
-
-- hosts: grafana
-  become: yes
-  gather_facts: yes
-  tasks:
-    - name: Include distribution variables for cloudalchemy.grafana
-      include_vars: "{{ appliances_repository_root }}/ansible/roles/cloudalchemy.grafana/vars/redhat.yml"
-    - import_role:
-        name: cloudalchemy.grafana
+    - name: kube prometheus stack
+      import_role:
+        name: kube_prometheus_stack
         tasks_from: install.yml
+      when: "'prometheus' in group_names"
 
 - name: Run post.yml hook
   vars:

ansible/monitoring.yml

@@ -28,11 +28,13 @@
     - import_role:
         name: filebeat
 
-- name: Deploy node_exporter
-  hosts: node_exporter
-  tags: node_exporter
+- name: Install monitoring helm chart
+  hosts: prometheus
+  become: true
+  tags: prometheus
   tasks:
-    - import_role: name=cloudalchemy.node_exporter
+    - ansible.builtin.import_role:
+        name: kube_prometheus_stack
 
 - name: Deploy OpenOndemand exporter
   hosts: openondemand
@@ -52,37 +54,3 @@
   tasks:
     - import_role:
         name: slurm_exporter
-
-- name: Setup core monitoring software
-  hosts: prometheus
-  tags: prometheus
-  tasks:
-    - name: Check for existing prometheus binaries
-      stat:
-        path: /usr/local/bin/{{ item }}
-      register: prometheus_binaries
-      loop:
-        - prometheus
-        - promtool
-    - name: Skip prometheus install if prometheus binaries exist and prometheus_version not defined
-      # i.e. if prometheus_version isn't defined we don't care, so use what's already there
-      set_fact:
-        prometheus_skip_install: "{{ false if prometheus_version is defined else true }}" 
-      when: "{{ (prometheus_binaries.results | map(attribute='stat') | map(attribute='exists')) + [prometheus_skip_install is not defined] }}"
-    - import_role:
-        name: cloudalchemy.prometheus
-
-- name: Deploy grafana
-  hosts: grafana
-  tags: grafana
-  tasks:
-    - assert:
-        that: vault_grafana_admin_password is defined
-        fail_msg: "Must define vault_grafana_admin_password - use `ansible-playbook generate-passwords.yml` to generate a set of passwords"
-    - include_role:
-        name: cloudalchemy.grafana
-      vars:
-        # We use internal roles to register the dashboards as the role does not support all options that we require.
-        grafana_dashboards: []
-    - import_role: # done in same play so it can use handlers from cloudalchemy.grafana
-        name: grafana-dashboards