too many unhealthy instances in the ring in a zone-aware configuration

[deferraz]

Hi community,

Considering a Grafana Mimir installation with 2.15 using zone-aware ingestion with 3 zones, if I have only pods from a zone off the ring stays online, if some pod from zone b reschedules together with zone a off or goes down from any reason the rings gets offline with that message of too many unhealthy instaces in the ring, I am having some problems understanding how can I get this part of the cluster more resilient, include more zones didn't help, if I have the same behaviour as said earlier the ingester ring is offline, what are the best practices here so the ring does not goes offline if more than one zone is rescheduled?