scheduling

# Scheduling restic Backups Run the backup script daily, on time, in the right timezone. Prefer a **systemd timer** (timezone-aware, handles missed runs, integrates with `OnFailure=` alerting). Use **cron** only on non-systemd hosts. Ready-to-paste units are in [references/systemd-units.md](references/systemd-units.md). **Do not enable the schedule until `verify-backup` passes** (first run + check + test-restore). Install the units now, enable them later. ## systemd timer (preferred) Two units in `/etc/systemd/system/`: `restic-backup.service` (oneshot, niced so it doesn't starve the host): ```ini [Unit] Description=restic backup to object storage (encrypted) Wants=network-online.target After=network-online.target docker.service [Service] Type=oneshot Nice=10 IOSchedulingClass=idle ExecStart=/usr/local/sbin/restic-backup.sh TimeoutStartSec=6h ``` `restic-backup.timer`: ```ini [Unit] Description=Daily restic backup [Timer] OnCalendar=*-*-* 03:00:00 Persistent=true RandomizedDelaySec=1800 [Install] WantedBy=timers.target ``` Reload and **verify the calendar resolves before enabling**: ```bash systemctl daemon-reload systemd-analyze calendar "*-*-* 03:00:00" --iterations=3 # prints the next 3 run times ``` ### Timezone systemd `OnCalendar` uses the **server's local time** by default — set the server timezone with `timedatectl set-timezone Europe/Kyiv`, or pin it per-timer (systemd ≥ 252): ```ini [Timer] OnCalendar=*-*-* 03:00:00 Timezone=Europe/Kyiv ``` `systemd-analyze calendar "*-*-* 03:00:00 Europe/Kyiv"` checks a zoned expression. `Persistent=true` runs a missed backup after downtime; `RandomizedDelaySec` spreads load and avoids hammering R2 at exactly the same second. ## cron fallback (non-systemd hosts) `/etc/cron.d/restic-backup` — set the timezone and prevent overlapping runs with `flock`: ```cron CRON_TZ=Europe/Kyiv 0 3 * * * root /usr/bin/flock -n /var/run/restic-backup.lock /usr/local/sbin/restic-backup.sh ``` `CRON_TZ` makes the schedule timezone-aware. `flock -n` ensures a long backup never overlaps the next run. cron's PATH is minimal — the script sets its own absolute PATH (see `backup-script`). ## Enable (only after verification) ```bash systemctl enable --now restic-backup.timer systemctl list-timers restic-backup.timer # shows NEXT run time ``` ## Gotchas - **Enable only after `verify-backup`** — never schedule an unverified backup. - **`daemon-reload` after editing units**, or systemd runs the old definition. - **`Persistent=true`** catches runs missed while the host was off. - **Avoid concurrent runs on one repo** — `flock` for cron; systemd oneshot won't overlap itself, but watch out for a manual run colliding with the timer (causes a repo lock → see `troubleshoot`). - **Timezone source** — system tz vs per-timer `Timezone=` vs cron `CRON_TZ`; pick one and verify with `systemd-analyze calendar`. - Schedule the periodic `check` (from `monitoring`) **off** the backup hour to avoid repo-lock contention. ## What this skill does NOT cover - The script itself → `backup-script` - First run + enabling gate → `verify-backup` - Failure alerts / `OnFailure=` wiring → `monitoring`