Using Pyrra with Google Managed Prometheus

[tobiasbrodersen]

On March 15 and onward, all newly created Google Kubernetes Engine standard clusters will have Managed Service for Prometheus in-cluster components enabled and deployed by default.

to me, this looks like a very attractive solution both cost and performance wise for many teams using Kubernetes who don't necessarily enjoy the complexity of running Prometheus themselves and something I could see many users adopt, including my own team... one blocker for us looks to be the lack of support with Pyrra.

For the managed version of this offering, GCP requires the user to define RecordingRules as Rules, ClusterRules and/or Global Rules which looks to be the well known PrometheusRuleSpec interface repacked in different types for users to define the scope of metrics that are to be "recorded".

I'm hoping to start a discussion on how the maintainers of Pyrra see the project fit together with this offering from GCP, it's something I'd love to work on, but I see multiple ways of approaching it e.g. running an "bridge" service that "translates" ServiceLevelObjective into the best matching GCP Monitoring rule, but that code could just as well live inside the project and be toggled on with a startup argument.

I'd love to hear anyones input and ideas :)
Thanks.

[ArthurSens]

Could pyrra generate be useful here? Maybe with some scripting during CI the rules could be uploaded?

I managed to upload rules to Grafana Cloud using 2 github actions: https://github.com/ArthurSens/pyrra-generate-action

[clearclaw]

Any chance of movement on this?