Skip to content

Configuration — Metrics

Buggregator has a new feature that collects metrics on how many events we receive. This feature lets you use tools like Prometheus and Grafana to check on event trends, find issues, and set up alerts.

This new feature will keep track of how many events Buggregator gets. It will sort these events by their types like sentry, monolog, var-dumper, ray, inspector, http-dump, profiler, and smtp.

How It Works

  1. Event Counting: Each time an event is received, it’s counted and sorted into its type.
  2. Metric Updating: The system updates the metrics to show the latest counts for each event type.

Prometheus Endpoint

Buggregator has a built-in Prometheus friendly endpoint that you can use to collect metrics. Metrics are available at 2112 port inside the container. To access the metrics outside the container, you need to expose the port.

bash
docker run --pull always \
  ... \
  -p 2112:2112 \
  ghcr.io/buggregator/server:latest

Note: Read more about server configuration here.

After starting the server, you can access the metrics at http://<server_address>:2112 address.

Metric Format

Here is what the metric format looks like:

plaintext
# HELP events The total number of received events.
# TYPE events counter
events{type="sentry"} 10
events{type="smtp"} 1

Each line shows how many events of a specific type we have received.

Grafana Integration

You can use these metrics to make dashboards in Grafana. This helps you see the data in a way that makes sense for you. We will provide steps on how to connect Buggregator’s metrics to Grafana.

Setting Up Grafana

  1. Add Prometheus as a Data Source: First, connect Prometheus (which has your Buggregator metrics) to Grafana.
  2. Create Dashboards: Next, create dashboards in Grafana to show the metrics. You can make graphs and charts that help you understand the event trends.