Marco Pracucci

_I think this is still valid_

> 1. There is already a limit of 140 samples per chunk for downsampled chunks here. I guess the performance should be similar to 120 samples? Yes, performances should be...

In the PR https://github.com/cortexproject/cortex/pull/3414 we've handled a similar case for `GetReplicationSetForOperation().` I'm wondering for which service and setup you're experiencing the issue. Assuming it's related to the ingesters ring (read/write...

You're right. What if we cancel only in case the error is a httpgrpc 5xx error or a non-httpgrpc error?

> Your suggestion would then make things worse, in this particular case. My suggestion was to cancel the rate-limiter reservation only in the case the distributor returns a 5xx, which...

I agree on having **no** new configuration option. Keep in mind our top 1 goal is to make Mimir easier to use. Less configuration, not more.

> HA_dedup;_all_samples_same_replica-10 I haven't looked at the code but I wasn't expecting this case to be slower. Have you considered an optimization for the happy path where we first analise...

The CHANGELOG has just been cut to prepare for the next Mimir release. Please rebase `main` and eventually move the CHANGELOG entry added / updated in this PR to the...

> How long should we wait before proceeding with this? _This issue originates from a conversation I had with Thomas (head security at Grafana) and Bryan._ My idea is to...

The reason why there's no auto-forget for ingesters is that if an unhealthy ingester is forgotten, then queries will succeed and return partial results instead of failing (Mimir was designed...