This message was deleted.

• historical node restart, non-persistent volumes causes slow start -> probably needs more replicas to be able to restart/lose pods without having unavailable segments • all historical has disk full? • historicals crashing due to OOM • missconfigured _tiers and not enough servers for all the given tiers •

Also sometimes segment will be not loaded and might be deleted from deep storage. You can figure out the segment name from historical logs and check in deep storage