The metrics endpoint

For an overview of other options for gathering statistics on Ambassador Edge Stack, see the Statistics and Monitoring overview.

Each Ambassador Edge Stack pod exposes statistics and metrics for that pod at http://{POD}:8877/metrics. The response is in the text-based Prometheus exposition format.

Understanding the statistics

The Prometheus exposition format includes special "HELP" lines that make the file self-documenting as to what specific statistics mean.

• envoy_*: See the Envoy documentation.
• ambassador_*:
  • ambassador_edge_stack_* (not present in Emissary-ingress):
    • ambassador_edge_stack_go_*: See [promethus.NewGoCollector()][].
    • ambassador_edge_stack_promhttp_* See promhttp.Handler().
    • ambassador_edge_stack_process_*: See [promethus.NewProcessCollector()][]..
  • ambassador_*_time_seconds (for * = one of aconf, diagnostics, econf, fetcher, ir, or reconfiguration): Gauges of how long the various core operations take in the diagd process.
  • ambassador_diagnostics_(errors|notices): The number of diagnostics errors and notices that would be shown in the diagnostics UI or the Edge Policy Console.
  • ambassador_diagnostics_info: Info about the Ambassador Edge Stack install; all information is presented in labels; the value of the Gauge is always "1".
  • ambassador_process_*: See prometheus_client.ProcessCollector.

Polling the :8877/metrics endpoint with Prometheus

To scrape metrics directly, follow the instructions for Monitoring with Prometheus and Grafana.

Using Grafana to visualize statistics gathered by Prometheus

Sample dashboard

We provide a sample Grafana dashboard that displays information collected by Prometheus from the :8877/metrics endpoint.

Additional Edge Stack latency metrics

Edge Stack provides some additional metrics around latency. Unlike the metrics from the above endpoint on port 8877, these metrics are collected and provided from a different component of Edge Stack. These additional metrics and the endpoint for them are opt-in. To enable them, set the environment variable ENABLE_PLUGIN_FILTER_METRICS to "true" on your Edge Stack deployment. By default, they will be available on the /metrics path on port 8878. The port can be configured using the PLUGIN_FILTER_METRICS_PORT environment variable.