fix raid

nextcloud/update_collabora.yml

@@ -1,6 +1,6 @@
-# nextcloud/update_collabora.yml
+# check_raid.yml
 
-- name: Update Collabora CODE on VM via Proxmox
+- name: Check Linux MD RAID health on VM via Proxmox
   hosts: linux_servers
   gather_facts: false
   become: true
@@ -8,33 +8,25 @@
   become_method: sudo
 
   vars:
-    # --- Connection to VM (provided by Semaphore env vars) ---
+    # VM connection (provided by Semaphore env vars)
     vm_ip:   "{{ lookup('env', 'VM_IP') }}"
     vm_user: "{{ lookup('env', 'VM_USER') }}"
     vm_pass: "{{ lookup('env', 'VM_PASS') }}"
     use_sudo: false
 
-    # --- Debug mode (controlled via Semaphore variable) ---
+    # Debug / retries
     DEBUG: "{{ lookup('env', 'DEBUG') | default(0) | int }}"
-    RETRIES: "{{ lookup('env', 'RETRIES') | default(25) | int }}"
+    RETRIES: "{{ lookup('env', 'RETRIES') | default(10) | int }}"
 
-    # --- Collabora specifics ---
+    # RAID device to check (e.g. md0, md1...)
-    collabora_debug_caps: true
+    raid_md_device: "{{ lookup('env', 'RAID_MD') | default('md0', true) }}"
-    collabora_caps_url: "https://collabora.martinfencl.eu/hosting/capabilities"
 
-    # Use the FULL Nextcloud stack compose file; only target the 'collabora' service inside it
+    # If 0 => fail when resync/recovery/reshape/check/repair is detected in /proc/mdstat
-    collabora_project: "nextcloud-collabora"
+    # If 1 => allow sync operations (still fails only on degraded [U_], [_U], etc.)
-    collabora_compose_file: "/data/compose/nextcloud/nextcloud-collabora.yml"
+    raid_allow_sync: "{{ lookup('env', 'RAID_ALLOW_SYNC') | default(1, true) | int }}"
-    collabora_service: "collabora"
 
-    # Docker command prefix (consistent behavior and quiet hints)
+    # SSH options
-    docker_prefix: "unalias docker 2>/dev/null || true; DOCKER_CLI_HINTS=0; command docker"
+    ssh_connect_timeout: 15
-
-    # Commands to run on the target VM (quiet outputs)
-    collabora_commands:
-      - "{{ docker_prefix }} pull -q collabora/code:latest >/dev/null"
-      - "{{ docker_prefix }} compose -p {{ collabora_project }} -f {{ collabora_compose_file }} pull {{ collabora_service }} >/dev/null"
-      - "{{ docker_prefix }} compose -p {{ collabora_project }} -f {{ collabora_compose_file }} up -d --no-deps --force-recreate {{ collabora_service }} >/dev/null"
 
   tasks:
     - name: Ensure sshpass is installed (for password-based SSH)  # English comments
@@ -43,7 +35,7 @@
         state: present
         update_cache: yes
 
-    - name: Run Collabora update commands on VM (via SSH)  # use SSHPASS env, hide item value
+    - name: Read /proc/mdstat from VM (via SSH)  # English comments
       ansible.builtin.command:
         argv:
           - sshpass
@@ -52,123 +44,74 @@
           - -o
           - StrictHostKeyChecking=no
           - -o
-          - ConnectTimeout=15
+          - "ConnectTimeout={{ ssh_connect_timeout }}"
           - "{{ vm_user }}@{{ vm_ip }}"
           - bash
           - -lc
-          - "{{ ('sudo ' if use_sudo else '') + item }}"
+          - "{{ ('sudo ' if use_sudo else '') + 'cat /proc/mdstat' }}"
       environment:
         SSHPASS: "{{ vm_pass }}"
-      loop: "{{ collabora_commands }}"
+      register: mdstat_raw
-      loop_control:
-        index_var: idx              # <-- capture loop index here
-        label: "cmd-{{ idx }}"      # <-- use idx instead of loop.index
-      register: collab_cmds
       changed_when: false
       no_log: "{{ DEBUG == 0 }}"
-
-    - name: Show outputs for each Collabora command
-      ansible.builtin.debug:
-        msg: |
-          CMD: {{ item.item }}
-          RC:  {{ item.rc }}
-          STDOUT:
-          {{ (item.stdout | default('')).strip() }}
-          STDERR:
-          {{ (item.stderr | default('')).strip() }}
-      loop: "{{ collab_cmds.results }}"
-      when: DEBUG == 1
-
-    - name: Fail play if any Collabora command failed  # also hide item label
-      ansible.builtin.assert:
-        that: "item.rc == 0"
-        fail_msg: "Collabora update failed on VM: {{ item.item }} (rc={{ item.rc }})"
-        success_msg: "All Collabora update commands succeeded."
-      loop: "{{ collab_cmds.results }}"
-      loop_control:
-        index_var: idx
-        label: "cmd-{{ idx }}"
-
-    # -------------------------
-    # Readiness checks (controller first, then VM fallback)
-    # -------------------------
-
-    - name: Collabora | Wait for capabilities (controller first)
-      ansible.builtin.uri:
-        url: "{{ collabora_caps_url }}"
-        method: GET
-        return_content: true
-        validate_certs: true
-        status_code: 200
-      register: caps_controller
-      delegate_to: localhost
-      run_once: true
       retries: "{{ RETRIES }}"
       delay: 2
-      until: caps_controller.status == 200
+      until: mdstat_raw.rc == 0
-      failed_when: false
-      changed_when: false
 
-    - name: Collabora | VM-side fetch (pure JSON via Python)  # use SSHPASS env here too
+    - name: Debug | Show raw /proc/mdstat  # English comments
-      ansible.builtin.command:
+      ansible.builtin.debug:
-        argv:
+        msg: "{{ mdstat_raw.stdout }}"
-          - sshpass
+      when: DEBUG == 1
-          - -e
-          - ssh
-          - -o
-          - StrictHostKeyChecking=no
-          - -o
-          - ConnectTimeout=15
-          - "{{ vm_user }}@{{ vm_ip }}"
-          - bash
-          - -lc
-          - |
-            python3 - <<'PY'
-            import json, urllib.request, sys
-            try:
-                with urllib.request.urlopen("{{ collabora_caps_url }}", timeout=15) as r:
-                    sys.stdout.write(r.read().decode())
-            except Exception:
-                pass
-            PY
-      environment:
-        SSHPASS: "{{ vm_pass }}"
-      register: caps_vm
-      changed_when: false
-      failed_when: false
-      when: caps_controller.status | default(0) != 200 or caps_controller.json is not defined
-      no_log: "{{ DEBUG == 0 }}"
 
-    - name: Collabora | Choose JSON (controller wins, else VM)
+    - name: Extract RAID status token for selected MD device  # English comments
       ansible.builtin.set_fact:
-        collab_caps_json: >-
+        raid_token: >-
           {{
-            (caps_controller.json
+            (mdstat_raw.stdout | regex_search(
-              if (caps_controller.status|default(0))==200 and (caps_controller.json is defined)
+              raid_md_device ~ '\\s*:\\s*active.*\\n\\s*\\d+\\s+blocks.*\\[[0-9]+/[0-9]+\\]\\s*\\[([U_]+)\\]',
-              else (
+              '\\1',
-                (caps_vm.stdout | default('') | trim | length > 0)
+              multiline=True
-                | ternary((caps_vm.stdout | trim | from_json), omit)
+            )) | default('')
-              )
-            )
           }}
-      failed_when: false
 
-    - name: Collabora | Print concise summary
+    - name: Detect sync operations in mdstat (resync/recovery/reshape/check/repair)  # English comments
+      ansible.builtin.set_fact:
+        raid_syncing: "{{ (mdstat_raw.stdout is search('resync|recovery|reshape|check|repair')) | bool }}"
+
+    - name: Compute degraded flag (underscore means missing member)  # English comments
+      ansible.builtin.set_fact:
+        raid_degraded: "{{ (raid_token | length > 0) and ('_' in raid_token) }}"
+
+    - name: Fail if MD device not found in /proc/mdstat  # English comments
+      ansible.builtin.assert:
+        that:
+          - raid_token | length > 0
+        fail_msg: "RAID device {{ raid_md_device }} was not found in /proc/mdstat on VM ({{ vm_ip }})."
+        success_msg: "RAID device {{ raid_md_device }} found in /proc/mdstat."
+      changed_when: false
+
+    - name: Fail if RAID is degraded (token contains '_')  # English comments
+      ansible.builtin.assert:
+        that:
+          - not raid_degraded
+        fail_msg: "RAID {{ raid_md_device }} is DEGRADED: token=[{{ raid_token }}] (expected all 'U')."
+        success_msg: "RAID {{ raid_md_device }} is OK: token=[{{ raid_token }}]."
+      changed_when: false
+
+    - name: Fail if RAID is syncing and syncing is not allowed  # English comments
+      ansible.builtin.assert:
+        that:
+          - (raid_allow_sync | int) == 1 or (not raid_syncing)
+        fail_msg: "RAID {{ raid_md_device }} is running a sync operation (resync/recovery/reshape/check/repair) and RAID_ALLOW_SYNC=0."
+        success_msg: "No sync operation detected (or RAID_ALLOW_SYNC=1)."
+      changed_when: false
+
+    - name: Print concise summary (debug)  # English comments
       ansible.builtin.debug:
         msg: >-
-          Collabora {{ collab_caps_json.productVersion | default('?') }}
+          RAID={{ raid_md_device }},
-          ({{ collab_caps_json.productName | default('?') }}),
+          token=[{{ raid_token }}],
-          convert-to.available={{ collab_caps_json['convert-to']['available'] | default('n/a') }},
+          degraded={{ raid_degraded }},
-          serverId={{ collab_caps_json.serverId | default('n/a') }}
+          syncing={{ raid_syncing }},
-      when: collab_caps_json is defined and DEBUG == 1
+          allow_sync={{ raid_allow_sync }}
-
+      when: DEBUG == 1
-    - name: Collabora | Capabilities unavailable (after retries)
-      ansible.builtin.debug:
-        msg: "Capabilities endpoint není dostupný ani po pokusech."
-      when: collab_caps_json is not defined and DEBUG == 1
-
-    # Optional full JSON (debug)
-    - name: Collabora | Full JSON (debug)
-      ansible.builtin.debug:
-        var: collab_caps_json
-      when: collabora_debug_caps and (collab_caps_json is defined) and DEBUG == 1