ServiceMonitor doens't work out of the box

[maxpain]

Hello. I've deployed the dragonfly-operator Helm chart with the following values:

serviceMonitor:
  enabled: true

But my VMAgent can't scrape metrics:

unexpected status code returned when scraping "http://10.244.2.242:8443/metrics": 400; expecting 200; response body: "Client sent an HTTP request to an HTTPS server.\n"

[Pothulapati]

Thanks for reporting!

I can try testing it out, but in the meanwhile we have something tested around metric collection of Dragonfly at https://www.dragonflydb.io/docs/managing-dragonfly/operator/prometheus-guide

[joggeli34]

We are facing the same problem.

As far as I see, the ServiceMonitor is targeting the port 8443 where as in the pod configuration the parameter --metrics-bind-address=127.0.0.1:8080 is set with a other port.

[jonathon2nd]

Also see: #242

[nikatar]

I have this problem too

To fix it, you need to change the logic:

1. start operator pod with -metrics-bind-address=0.0.0.0:8080
2. this port(8080) must set in ports of deployment
3. serviceMonitor is targeted to this port

And why don't you use podMonitor instead serviceMonitor? This operator deployment doesn't seem to require the service, as it doesn't accept traffic.