Skip to content

Scheduled Jobs

EMBRACE has two scheduled jobs: - Refresh: the daily Emby watch history refresh. It can be scheduled with a systemd user timer (used on optiplex01) or with cron; both run the same Python script. - Picks: the nightly recommendation playlists, which run straight after the refresh (systemd only).

systemd timer (optiplex01)

Unit files live in scripts/systemd/ and assume the repo is at ~/src/EMBRACE with a venv at ~/src/EMBRACE/venv. A timer catches up a run missed while the machine was off (Persistent=true) and never runs two refreshes at once.

mkdir -p ~/.config/systemd/user
ln -sf ~/src/EMBRACE/scripts/systemd/embrace-refresh.{service,timer} ~/.config/systemd/user/
systemctl --user daemon-reload
systemctl --user enable --now embrace-refresh.timer
loginctl enable-linger "$USER"     # keep user timers running without a login session

systemctl --user start embrace-refresh.service   # run once now
journalctl --user -u embrace-refresh -n 50       # logs
systemctl --user list-timers embrace-refresh.timer

The canonical SQLite database is the one on the machine running the refresh (optiplex01: ~/src/EMBRACE/sqlite_db/). To work with a current copy elsewhere:

scp optiplex01:src/EMBRACE/sqlite_db/EMBRACE_db.db sqlite_db/

Daily picks

embrace-picks.service runs scripts/picks/daily_picks.py (see Daily picks). - Trigger: it has no timer of its own. Enabling it links it to the refresh, which pulls it in on every run. - Order: it starts once the refresh has finished, even if the refresh failed, since yesterday's data still makes fair picks.

systemctl --user enable ~/src/EMBRACE/scripts/systemd/embrace-picks.service   # links it + hooks it to the refresh
systemctl --user start embrace-picks.service      # run once now
journalctl --user -u embrace-picks -n 50

The job needs the IMDB MySQL database (or its genre cache in data/cache/). For the collaborative signal, it also needs MovieLens in data/movielens/.

Cron

Emby Watch History Refresh

TODO: add environment variable for custom discord webhook

Located under: scripts/cron/emby_refresh_watch_hist.sh

The purpose of this cron job is to refresh the Emby watch history database daily at 00:00.

This script has three modes, you can also use the "--help" flag to see these options: - install_refresh_cron.sh - install or update the cron job - install_refresh_cron.sh --remove - remove the cron job - install_refresh_cron.sh --run - run the job once now (same command cron would run)

Running this script will add a cron job to the executing machine which does the following: 1. Copies the current SQLite database to sqlite_db/backups, with the current date + time (Australia/Melbourne timezone) and ".backup" appended 2. Makes a working copy of the database (<SQLITE_DB_NAME>.building) and refreshes it: TMDB genres (optional), library metadata, TMDB title details for new or stale titles (optional - keywords, cast/crew, descriptions and TMDB recommendations), Emby user item states (optional - played flags), Jellyseerr requests (optional, when configured), new watch events, then rebuilds sessions and stats 3. Replaces the live database with the working copy only if every step succeeded - on failure the working copy is discarded, the live database is untouched, and a failure message is sent to Discord

Watch events are added incrementally rather than rebuilt, because Playback Reporting only keeps ~90 days of history.

The mentioned python script is located under scripts/sqlite/emby_refresh_watch_hist.py.