Adopt for bare-metal setups

[GiedriusS]

Hello! Thank you for your project. I was thinking about how much effort it would be to make this suitable for bare-metal setups as well. Essentially, it should perform all of the same things except for the automation of scaling down/up. The goal is to have an equal load over all Prometheus instances. Do you think this would make sense for your project? If yes maybe I could contribute.

[RayHuangCN]

Hi 😊 , thank you for your advice! We already support a simple static shards way for bare-metal now.
User can provide a shards config for Coordinator to indicate shards information.

• add --shard.type=static and --shard.static-file=xxx.yaml for Coordinator
• add --config.file="" for Sidecar to get config file from Coordiantor (no local config file needed)

but this feature is kind of weak now

• shards can only use the same max series
• scaling down/up event is unavailable.

I think that we can support http way, which use a shards server for fetch shards information, and deal with scaling down/up event.

[GiedriusS]

I see. I'll try it out and come back with any comments/suggestions. 👍