Problem
For metrics, we have been patching the prometheus configuration installed with serving. See here. This is kind of a hack. We can start using the prometheus operator across Knative to avoid such issues.
Persona:
Contributor
Exit Criteria
Documentation on how to set up the operator is available in knative/docs. There is no need to patch configurations for scrapping additional resources anymore.
Additional context (optional)
This was presented sometime back by @n3wscott in the Observability WG and is something that can benefit Knative in general.
/cc @Abd4llA @sayanh
fyi @Harwayne @mdemirhan
/project Observability To do
@sayanh @Abd4llA so that you have it on your radar in case you have the cycles
+1 for this, it would help test https://github.com/knative/eventing-contrib/issues/612 once its done.
/assign @marcobebway
/project Observability In progress
@marcobebway: You must be a member of the knative/eventing github team to set the project and column.
In response to this:
/project Observability In progress
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
/project Observability In progress
@Harwayne: You have successfully moved the project card for this issue to column In progress (ID 6212148).
In response to this:
/project Observability In progress
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.
fyi @mdemirhan
The PR is created on the knative/observability repository https://github.com/knative/observability/pull/52.
@marcobebway thanks for doing this! Quick question, can we remove the ugly patching we are doing to the config maps in eventing with this? If so, would it be possible to update the docs there?
@nachocano, I agree with you about removing the scrape config patching from eventing and use the Prometheus operator and service monitors instead! Please let me think about how we can move this forward!
Also, I think that this might involve how the monitoring config is bundled as part of the Knative release, what do you think?