Julien

Thanks!

We have discussed this during our bug scrub. We found out that it is confusing that this pull request does not explain how to do capturing groups. Maybe the sentence...

Can we maybe split the named capture group into a second exemple with the explanation?

Thanks!

In order to better understand the issue, could you provide more information about the status of the connections/sockets and any error messages you may be seeing? It's also possible that...

for Prometheus 3.0.

Thanks for your suggestion. That would make remote write a lot less efficient. It seems like something to tackle at the load balancer level, or the store proxies could itself...

@csmarchbanks well it turns out we have service discovery already, why would we reinvent something? could we reuse service discovery, like we have for alertmanagers, and reshard on targets change?

> If we used service discovery I imagine we would have to manually implement health checks/retries/etc, which we don't have to do for Alertmanager since we send to all instances...

But for the rest I agree that it is a different usecase that we could manage differently.