feat: add serviceMonitor for apisix-gateway

[mstefany]

Metrics' port & URL is defined here (port 9091 & URL /apisix/prometheus/metrics).

And it doesn't match what's in the serviceMonitor. TargetPort should be prometheus, a.k.a. 9091, and path is not specified which defaults to /metrics which is incorrect.

Fact that it's not working in default configuration is visible in my logs:

apisix-ingress-controller-6b4d5cc778-f582v apisix 10.254.82.207 - - [17/Jan/2024:11:00:22 +0000] 10.254.83.238:9080 "GET /metrics HTTP/1.1" 404 47 0.000 "-" "Prometheus/2.48.1" - - - "http://10.254.83.238:9080"

[mstefany]

Please, is it possible to have this reviewed and approved?

[zll600]

ping @AlinsRan @Revolyssup

[zll600]

ping @AlinsRan @Revolyssup

[flassagn]

Guys any feedback about the review ? We would like to enable the ingress controller monitoring ! 🙏

[Revolyssup]

Guys any feedback about the review ? We would like to enable the ingress controller monitoring ! 🙏

I will take a look at this tomorrow and review. Thanks for reminding.

[mstefany]

I pushed the updated version, so that linter will be happy this time.

[flassagn]

Hi @Revolyssup,
The PR needs two approval. Could you assign another maintainer here ? I really would like to scrape ingress controller metrics ! 🙏

[mstefany]

@AlinsRan @Revolyssup I updated the PR:

• because both apisix-ingress-controller and apisix containers have ports named http, original ServiceMonitor would scrape both containers (if apisix-gateway is enabled), and cause error every 15 seconds, because there are no metrics on http port 9080 of gateway at /metrics URL. For this reason I am renaming http port of apisix-gateway-controller to http-controller.
• enabling ServiceMonitor of gateway is now done separately in .Values.gateway.serviceMonitor, and it also provisions separate ServiceMonitor object to scrape such metrics

Please, review the code if this approach is acceptable. I have updated README.md with helm-docs so there shouldn't be linter errors.