Deployment¶

This guide shows how to run SUISA Sendemeldung in production so that the monthly report is generated and sent automatically on a schedule.

Supported deployment models¶

Model	Recommended for
systemd timer	Self-hosted Linux servers (bare metal or VM)
Container (one-shot)	Any host with rootless Podman
Cron job	Minimal setups without systemd

systemd timer¶

The etc/systemd/ directory in this repository ships a ready-to-use template unit pair.

Quick setup¶

# 1. Copy configuration
sudo mkdir -p /etc/suisa_sendemeldung
sudo cp etc/suisa_sendemeldung.toml /etc/suisa_sendemeldung/production.toml
# edit the file and fill in your credentials
sudo $EDITOR /etc/suisa_sendemeldung/production.toml

# 2. Install units
sudo cp etc/systemd/suisa_sendemeldung@.service /etc/systemd/system/
sudo cp etc/systemd/suisa_sendemeldung@.timer   /etc/systemd/system/

# 3. Enable and start
sudo systemctl daemon-reload
sudo systemctl enable --now 'suisa_sendemeldung@production.timer'

# 4. Verify
systemctl list-timers suisa_sendemeldung@production.timer

How the units work¶

suisa_sendemeldung@.service is a template service that:

Pulls the latest container image before each run.
Mounts /etc/suisa_sendemeldung/%i.toml as /etc/suisa_sendemeldung.toml inside the container (where %i is the instance name — production in the example above).
Runs the container as a one-shot job and exits when done.

suisa_sendemeldung@.timer triggers the service on the 14th of each month at 09:00. RaBe runs the script on the 1st and 14th; adjust the OnCalendar value to suit your schedule.

Multiple stations

Because the units are template units you can run several instances in parallel, one per station configuration file:

systemctl enable --now suisa_sendemeldung@station-a.timer
systemctl enable --now suisa_sendemeldung@station-b.timer

Customising the schedule¶

Edit the timer's OnCalendar value to change when the report runs:

[Timer]
# Run on the 1st and 14th of every month at 09:00
OnCalendar=
OnCalendar=*-*-1,14 09:00:00
Persistent=true

After editing, reload the daemon and restart the timer:

sudo systemctl daemon-reload
sudo systemctl restart suisa_sendemeldung@production.timer

Container (one-shot)¶

Run the container once for a specific month using rootless Podman:

podman run --rm \
  -v /etc/suisa_sendemeldung/production.toml:/etc/suisa_sendemeldung.toml:ro \
  ghcr.io/radiorabe/suisasendemeldung:latest \
  suisa_sendemeldung --last-month

Docker

Docker works as a drop-in replacement — substitute docker for podman in any of the commands above. Rootless Podman is strongly recommended.

All configuration can also be passed as environment variables without a mounted config file:

podman run --rm \
  -e SENDEMELDUNG_ACR_BEARER_TOKEN=ey... \
  -e SENDEMELDUNG_ACR_STREAM_ID=a-bcdefgh \
  -e SENDEMELDUNG_ACR_PROJECT_ID=1234 \
  -e SENDEMELDUNG_OUTPUT=email \
  -e SENDEMELDUNG_EMAIL_SERVER=smtp.example.org \
  -e SENDEMELDUNG_EMAIL_SENDER=report@example.org \
  -e SENDEMELDUNG_EMAIL_TO=sendemeldung@suisa.ch \
  -e SENDEMELDUNG_EMAIL_PASSWORD=s3cr3t \
  -e SENDEMELDUNG_DATE_LAST_MONTH=true \
  ghcr.io/radiorabe/suisasendemeldung:latest

Cron job¶

If systemd is not available, use cron:

# Run on the 14th of every month at 09:00
0 9 14 * * podman run --rm \
  -v /etc/suisa_sendemeldung/production.toml:/etc/suisa_sendemeldung.toml:ro \
  ghcr.io/radiorabe/suisasendemeldung:latest \
  suisa_sendemeldung --last-month

Monitoring¶

The systemd service file contains a commented-out ExecStartPost line that sends a Zabbix heartbeat after a successful run. Uncomment and adapt it to your monitoring stack:

ExecStartPost=-/bin/sh -c 'zabbix_sender \
  -c /etc/zabbix/zabbix_agent2.conf \
  -s sendemeldung.example.org \
  -k rabe.suisa_sendemeldung.run.success \
  -o "$$(date +%%s)"'

Double % in systemd units

systemd unit files use %% to produce a literal %. If you adapt this command for a plain shell script, replace %%s with %s.