-
Story
-
Resolution: Done
-
Major
-
None
-
None
To achieve horizontal scalability of our scraping layer, we would like to distribute Service/PodMonitors and any other targets to multiple Scrapers dynamically.
This is similar to what Otel is doing: https://docs.google.com/document/d/13Gcu5SlbgjrsQJQUuZAjdQo1MOQA76Yji3oX8yHh-p8/edit#heading=h.tq4cuijg0dif so we might want to join forces.
AC:
- Design (and optionally PoC) for a solution that allows us to run Prometheus Operator in a mode that allows us to set a group of Prometheus-es that take care of a group of Service/Pod Monitors. Such group will dynamically scale up and down to match the number of targets per instance.
- Check if we can reuse Otel work (or Otel can reuse our work)
It's worth to think about setting some smarter sharding e.g per node in future too.