service nacos-cs routed the request to an unavailable instance

[fyuan1316]

release 2.1.1

description:

The nacos cluster is deployed with 3 instances and a power failure causes an abnormal status of 1 instance.
The error is returned by the client requesting /nacos/v1/ns/instance/beat with the configured nacos address
server-addr: nacos-cs。

---

I think in this case if the readinessProbe check is configured in the statefulset, the unavailable instances will be removed from the service load balancing and this should solve the problem.

I see that there was a discussion #186 about this, and the restart issue described here should have been resolved with the startupProbe. So, if there are no other issues, I think we can add readinessProbe back in.

[fyuan1316]

@paderlol
WDYT ?

[wilsonwu]

Good suggestion.