Monitoring

Every Sourcegull process serves Prometheus metrics at /metrics on its internal listener (:6060). Sourcegull ships alerting rules and Grafana dashboards for them, generated from crates/monitoring. Load them into your own Prometheus and Grafana:

• alerts.md: every alert, with next steps. Rules file: prometheus/sourcegull-alerts.yml in the Compose bundle.
• dashboards.md: one dashboard per service. JSON files: grafana/dashboards/*.json in the Compose bundle.

The dashboards and alerts select series by a role label: web, worker, repo or blob for a process running that role, or all for one running every role (sourcegull supervise, --roles=all). Set it in the scrape config.

#Helm

The chart creates the ServiceMonitor (metrics.serviceMonitor.enabled=true, sets role), the PrometheusRule (metrics.prometheusRule.enabled=true) and the dashboard ConfigMaps for the Grafana sidecar (metrics.grafanaDashboards.enabled=true). See the chart README.

#Docker Compose

prometheus/prometheus.yml in the bundle is an example scrape config for both Compose files.

The internal listener is not published: it also serves the blob API of a process running the blob role (compose.yaml, and the blob service of compose.production.yaml), which has no authentication. A Prometheus in a container on the same host can join the Compose network and scrape sourcegull:6060 (or web:6060, … in the production layout):

# In your Prometheus' Compose file.
services:
  prometheus:
    networks: [default, sourcegull]
networks:
  sourcegull:
    name: sourcegull_default
    external: true

For a Prometheus elsewhere, publish the port on an address that only it can reach, with a compose.override.yaml next to compose.yaml (Compose reads it with the default file names):

services:
  sourcegull:
    ports:
      - "10.0.0.5:6060:6060"   # a private address of this host

With compose.production.yaml, put the override in its own file and add it with -f (or to COMPOSE_FILE in .env, separated by :), and publish each service on its own host port.

Then:

1. Add the scrape job with the role label (as in the example), and prometheus/sourcegull-alerts.yml under rule_files.
2. In Grafana, import each grafana/dashboards/*.json (Dashboards → New → Import) and pick your Prometheus data source. Or provision the directory with a dashboard provider.