Proposal

Over the years, the project accumulated 29 service discovery mechanisms. While some SDs are widely used (e.g. Kubernetes, EC2, GCP, Azure, ...), the project faces a couple of issues:

• No clear ownership for some SDs. For instance, nobody part of the Prometheus team has direct experience with Azure (dev summit notes).
• "Critical" service discoveries like Kubernetes sometimes lack timely reviews.
• Very limited end-to-end testing (e.g. manual only).
• A high number of SDs means bigger binary size because of SDK dependencies.

The goals of this issue are to:

• Identify official maintainers for each SD.
• Evaluate whether we want/need to deprecate some of them (since Prometheus 3.0 is around the corner, it might be the right time).

I've put together a spreadsheet to detail the current state.