@atoulme My idea is to add an UpdateInterval field to the Config struct, which will also be added to the CLI flag, could I take over this?

This issue has not been triaged by the project, but yes, that's what I think should be done. Thanks!

Is there any specific reason you'd like to change this value? It only exists for rate limiting changes to the set of collectors. Any such change triggers a target reallocation, which can be expensive. We rate limit changes to targets themselves in the same way, and the interval is also not configurable there.

I'm not opposed to making it configurable, but I'd like to understand why that's necessary first.

This minimum is too small and we are seeing cases where the operator runs aground of the throttling limits of the API server. In general, we want all the knobs available to raise those limits for large customers. This is just one such occurrence, more will come up as we investigate.

This minimum is too small and we are seeing cases where the operator runs aground of the throttling limits of the API server.

I'm not sure I understand. This value doesn't have any effect on how target allocator talks to the API Server - in this case, there's a watch on collector Pods, and it receives all the updates. The only thing this value affects is how often targets are reallocated as a result of changes to the active collector set.