Aller au contenu principal
Version: v4

Standalone

Standalone mode runs only the Python app. You must provide your own InfluxDB 3 instance (and Grafana, if you want dashboards) to build the stack described in the Design section.

astuce

If you don't already run InfluxDB and Grafana, prefer the Docker Compose stack.

Requirements​

  • Windows, Linux or macOS
  • Python 3.10+
  • uv (used to manage dependencies)
  • A reachable InfluxDB 3 Core instance and an API token

Preparing steps​

  1. Download the source code from the latest release and extract it.

  2. Install the dependencies from the app folder (uv creates an isolated .venv from the lockfile):

cd app
uv sync
  1. Write a config.yaml (manual mode) — or a settings.yaml + events/ directory (events mode) — following the configuration section, in the SuiviBourse config directory:
  • ~/.config/SuiviBourse/
info

Manual mode resolves the config location through the confuse library, so other standard search paths are supported too.

Running the app​

From the app folder, provide at least the InfluxDB token and run:

INFLUXDB_HOST=http://localhost:8181 \
INFLUXDB_TOKEN=<your-token> \
uv run gunicorn -c src/gunicorn.conf.py 'web:create_app()'
info

The app is served by gunicorn: the web API and the scheduler share a single process, so there is one command and one process for both. gunicorn.conf.py is the boot sequence — it binds the web port (SB_WEB_PORT, default 8080) and the legacy /metrics port, and refuses to start with more than one worker. Raise threads, never workers: each extra worker would run a second scraper and duplicate every write.

For production, choose one of the following methods.

Replace <installation_path> with the folder where you extracted the project, and <your-token> with your InfluxDB API token.

  1. Create a service unit file in /etc/systemd/system/suivi-bourse.service (replace <user>/<group> and the token):

    [Unit]
    Description=SuiviBourse
    After=network-online.target

    [Service]
    Type=simple
    Restart=always
    WorkingDirectory=<installation_path>/app
    ExecStart=uv run gunicorn -c src/gunicorn.conf.py 'web:create_app()'
    User=<user>
    Group=<group>
    Environment=PYTHONUNBUFFERED=1
    Environment=INFLUXDB_HOST=http://localhost:8181
    Environment=INFLUXDB_TOKEN=<your-token>
    Environment=INFLUXDB_DATABASE=suivi_bourse
    SyslogIdentifier=suivi-bourse

    [Install]
    WantedBy=default.target
    remarque

    ExecStart must use an absolute path to the uv binary (e.g. /usr/local/bin/uv) — run which uv to find it.

  2. Reload systemd:

    systemctl daemon-reload

  3. Enable the service at boot:

    systemctl enable suivi-bourse

  4. Start the service:

    systemctl start suivi-bourse

Logs go to /var/log/syslog; check the status with systemctl status suivi-bourse.

Environment variables​

At minimum you must set the InfluxDB connection variables:

VariableDefaultDescription
INFLUXDB_HOSThttp://influxdb:8181InfluxDB 3 host URL
INFLUXDB_TOKEN(required)InfluxDB API token
INFLUXDB_DATABASEsuivi_bourseInfluxDB database name
SB_REGULAR_INTERVAL120Poll interval (seconds) while a symbol's market is open (closed markets sleep to the next open)
SB_SCRAPING_INTERVAL(deprecated)Deprecated fallback for SB_REGULAR_INTERVAL (honored with a warning)
SB_WEB_PORT8080Port for the web API and its /health endpoint
SB_METRICS_PORT8081Port for the legacy Prometheus /metrics endpoint
LOG_LEVELINFOLogging level

See the configuration overview for the full list.

Grafana dashboard​

Point Grafana at your InfluxDB instance using an InfluxDB datasource in SQL query mode, then import the provisioned dashboard from the repository:

An example datasource definition is available in grafana_provisioning/datasources/all.yml.

Prefer Prometheus?

If you still rely on a Prometheus/Grafana stack, the app keeps exposing a legacy /metrics endpoint. See Legacy Prometheus endpoint.