ZFS: scrub: cron --> systemd-timer (#341)

Co-authored-by: Michael Grote <michael.grote@posteo.de>
Reviewed-on: mg/ansible#341
Co-authored-by: mg <michael.grote@posteo.de>
Co-committed-by: mg <michael.grote@posteo.de>

group_vars/all.yml

@@ -21,7 +21,7 @@
         # https://forum.restic.net/t/exclude-syntax-confusion/1531/12
   restic_mount_timeout: "10 min"
   restic_failure_delay: "30 s"
-  restic_schedule: "0/6:00"
+  restic_schedule: "0/6:00" # alle 6 Stunden
   restic_folders_to_backup: "/" # --one-file-system ist gesetzt, also werden weitere Dateisysteme nicht eigeschlossen, es sei denn sie werden hier explizit angegeben; https://restic.readthedocs.io/en/latest/040_backup.html#excluding-files
   restic_repository: "//fileserver2.grote.lan/restic"
   restic_repository_password: "{{ lookup('keepass', 'restic_repository_password', 'password') }}"

host_vars/pve2-test2.grote.lan.yml

@@ -42,14 +42,6 @@
   zfs_extra_arc_max_size: "2147483648" # 2GB in Bytes
   zfs_extra_zfs_pools:
     - name: "rpool"
-      cron_minutes_zfs_scrub: "5"
-      cron_hour_zfs_scrub: "23"
-      cron_day_of_month_zfs_scrub: "14"
-      cron_day_of_week_zfs_scrub: "*"
-      cron_month_zfs_scrub: "*/2"
+      systemd_timer_schedule: "*-01,04,07,10-01 23:00"
     - name: "hdd_vm_mirror"
-      cron_minutes_zfs_scrub: "10"
-      cron_hour_zfs_scrub: "23"
-      cron_day_of_month_zfs_scrub: "14"
-      cron_day_of_week_zfs_scrub: "*"
-      cron_month_zfs_scrub: "*/2"
+      systemd_timer_schedule: "*-01,04,07,10-01 23:00"

host_vars/pve2.grote.lan.yml

@@ -177,17 +177,9 @@
   zfs_extra_arc_max_size: "8589934592" # 8GB in Bytes
   zfs_extra_zfs_pools:
     - name: "rpool"
-      cron_minutes_zfs_scrub: "0"
-      cron_hour_zfs_scrub: "23"
-      cron_day_of_month_zfs_scrub: "14"
-      cron_day_of_week_zfs_scrub: "*"
-      cron_month_zfs_scrub: "*/2"
+      systemd_timer_schedule: "*-01,04,07,10-01 23:00" # jeden ersten eines jeden Quartals
     - name: "hdd_data_raidz"
-      cron_minutes_zfs_scrub: "0"
-      cron_hour_zfs_scrub: "23"
-      cron_day_of_month_zfs_scrub: "14"
-      cron_day_of_week_zfs_scrub: "*"
-      cron_month_zfs_scrub: "*/2"
+      systemd_timer_schedule: "*-01,04,07,10-01 23:00"
 
   ### mgrote.zfs_sanoid
   sanoid_datasets:

roles/mgrote.restic/defaults/main.yml

@@ -13,11 +13,11 @@
         **/**cache***/**
         **/**Cache***/**
         **/**AppData***/**
-  # timeout for cifs mount; systemd notation
+  # timeout for cifs mount; systemd notation(systemd-analyze calendar "Sun *-0/2-* 23:00")
   restic_mount_timeout: "10 min"
-  # delay for restartung task; systemd notation
+  # delay for restartung task; systemd notation(systemd-analyze calendar "Sun *-0/2-* 23:00")
   restic_failure_delay: "30 s"
-  # when should restic run; systemd notation
+  # when should restic run; systemd notation(systemd-analyze calendar "Sun *-0/2-* 23:00")
   restic_schedule: "*:0/2"
   # welche ordner sollen gesichert werden
   restic_folders_to_backup: "/usr/local /etc /root /var/www /home"

roles/mgrote.zfs_scrub/README.md

@@ -1,13 +1,12 @@
 ## mgrote.zfs_scrub
 
 ### Beschreibung
-Richtet regelmaessige Scrubs(jeden Sonntag) ein.
+Richtet regelmaessige Scrubs mit systemd-timer ein.
 
 
 ### getestet auf
-- [x] ProxMox 6.1
+- [x] ProxMox 7*
 
 ### Variablen + Defaults
 - see [defaults](./defaults/main.yml)
 - Variablen für ``mgrote.zfs_health/trim/scrub/zed/arc_mem/`` sind zusammengefasst unter zfs_extra_*
-- see [default-filter](./tasks/main.yml)

roles/mgrote.zfs_scrub/tasks/main.yml

@@ -1,20 +1,84 @@
 ---
-  - name: remove system scrub job #https://forum.proxmox.com/threads/script-sequential-zfs-scrub-for-cron.25124/
+  - name: template unit-template
+    become: true
+    ansible.builtin.template:
+      src: zpool-scrub@.service.j2
+      dest: /etc/systemd/system/zpool-scrub@.service
+      owner: root
+      group: root
+      mode: 0644
+    register: template_unit
+
+  - name: template timer
+    become: true
+    ansible.builtin.template:
+      src: zpool-scrub@.timer.j2
+      dest: "/etc/systemd/system/zpool-scrub@{{ item.name }}.timer"
+      owner: root
+      group: root
+      mode: 0644
+    loop: "{{ zfs_extra_zfs_pools }}"
+    register: template_timer
+
+  - name: template mail.service
+    become: true
+    ansible.builtin.template:
+      src: zpool-scrub-mail.service.j2
+      dest: "/etc/systemd/system/zpool-scrub-mail.service"
+      owner: root
+      group: root
+      mode: 0644
+    register: template_mail
+
+  - name: systemctl daemon-reload
+    become: yes
+    ansible.builtin.systemd:
+      daemon_reload: yes
+    when:
+      - template_unit.changed
+      - template_timer.changed
+
+  - name: systemctl enable units - timer
+    become: yes
+    ansible.builtin.systemd:
+      name: "zpool-scrub@{{ item.name }}.timer"
+      enabled: yes
+      masked: no
+    loop: "{{ zfs_extra_zfs_pools }}"
+    when:
+      - template_unit.changed
+      - template_timer.changed
+
+  - name: systemctl enable units - mail
+    become: yes
+    ansible.builtin.systemd:
+      name: "zpool-scrub-mail.service"
+      enabled: yes
+      masked: no
+    when:
+      - template_mail
+
+  - name: systemctl start units
+    become: yes
+    ansible.builtin.systemd:
+      name: "zpool-scrub@{{ item.name }}.timer"
+      state: restarted
+      enabled: yes
+    loop: "{{ zfs_extra_zfs_pools }}"
+    when:
+      - template_unit.changed
+      - template_timer.changed
+
+  - name: remove old cronjobs
+    become: yes
+    ansible.builtin.cron:
+      name: zfs-scrub - "{{ item.name }}"
+      state: absent
+    with_items: "{{ zfs_extra_zfs_pools }}"
+
+  - name: remove proxmox system scrub job #https://forum.proxmox.com/threads/script-sequential-zfs-scrub-for-cron.25124/
     become: true
     ansible.builtin.lineinfile:
       path: /etc/cron.d/zfsutils-linux
       state: absent
       line: '24 0 8-14 * * root [ $(date +\%w) -eq 0 ] && [ -x /usr/lib/zfs-linux/scrub ] && /usr/lib/zfs-linux/scrub'
-
-  - name: add cronjob "zfs-scrub"
-    become: yes
-    ansible.builtin.cron:
-      name: zfs-scrub - "{{ item.name }}"
-      state: present
-      job: "{{ zfs_extra_path_zpool }} scrub {{ item.name }}"
-      month: "{{ item.cron_month_zfs_scrub | default('*') }}"
-      weekday: "{{ item.cron_day_of_week_zfs_scrub | default('7') }}"
-      day: "{{ item.cron_day_of_month_zfs_scrub | default('*/21') }}"
-      hour: "{{ item.cron_hour_zfs_scrub | default('23') }}"
-      minute: "{{ item.cron_minutes_zfs_scrub | default('30') }}"
-    with_items: "{{ zfs_extra_zfs_pools }}"

roles/mgrote.zfs_scrub/templates/zpool-scrub-mail.service.j2

@@ -0,0 +1,8 @@
+{{ file_header | default () }}
+
+[Unit]
+Description=Send a Mail in case of an error in zpool-scrub@.service.
+
+[Service]
+Type=oneshot
+ExecStart=/bin/bash -c '{{ zfs_extra_path_zpool }} status | mail -s "[ERROR] zpool scrub - %H" {{ empfaenger_mail }}'

roles/mgrote.zfs_scrub/templates/zpool-scrub@.service.j2

@@ -0,0 +1,13 @@
+{{ file_header | default () }}
+
+[Unit]
+Description=Scrub ZFS pool "%i"
+Requires=zfs.target
+After=zfs.target
+OnFailure=zpool-scrub-mail.service
+
+[Service]
+Type=simple
+ExecStartPre=-{{ zfs_extra_path_zpool }} scrub -s %i
+ExecStart={{ zfs_extra_path_zpool }} scrub -w %i
+ExecStop={{ zfs_extra_path_zpool }} scrub -s %i

roles/mgrote.zfs_scrub/templates/zpool-scrub@.timer.j2

@@ -0,0 +1,13 @@
+{{ file_header | default () }}
+
+[Unit]
+Description=Scrub ZFS pool "%i"
+
+[Timer]
+OnCalendar={{ item.systemd_timer_schedule }}
+Persistent=true
+Unit=zpool-scrub@%i.service
+RandomizedDelaySec=30 min
+
+[Install]
+WantedBy=timers.target