Feature Request: Support Prometheus / Openmetrics

[Tuxdude]

It would be great to get a /metrics endpoint (or an alternative endpoint if /<something> always means the user something) to allow the opengist job to be monitored using Prometheus or other metrics collector that supports the Openmetrics format.

Even if we can expose just an up metric to begin with, that would be a great start to probe the health of through Prometheus. Other metrics to export can be decided and added subsequently.

[thomiceli]

/healthcheck already exists. The idea of metrics is interesting indeed

[Tuxdude]

/healthcheck already exists. The idea of metrics is interesting indeed

Unfortunately the response for /healtcheck doesn't comply with the openmetrics format for a minimum /metrics page.

Prometheus currently returns the following scrape error if I make it scrape the /healthcheck endpoint of opengist.

expected equal, got ":" ("INVALID") while parsing: "{\"database\":"

The actual output from opengist /healthcheck is:

{
  "database": "ok",
  "opengist": "ok",
  "time": "2024-09-10T10:28:09-07:00"
}

Is it possible to expose a /metrics endpoint that just returns an empty response body with 200 status code? That would satisfy the minimum requirements to keep prometheus and other openmetrics based scrapers happy.