I have three nodes in a cluster, and have enabled auto-eviction. I’ve seen that happen once (probably because of an issue on the host). Once it happened, the evicted node stayed up and appeared to process read operations ok, but bounced modification operations - which I think is as intended.
I’m using haproxy as a load balancer in front of these nodes, and in that environment it would be better (for a lot of applications) if the evicted node went down, so that clients were redirected to working nodes.
Is there an automated way of doing this? Obviously I could write a script which looks at the node state and shut it down, but I don’t want to reinvent wheels.