Errors on clustering when switching nodepools

[kuzmik]

We ran a failover today that moved proxysql to a different nodepool, and the new one was not set to autoscale. This lead to having 2 pods be unscheduled, but the agent was still trying to cluster with those pods.

I thought there was code to prevent clustering if the pod doesn't have an IP yet, I will check on that.

Alternatively something is screwy with k8s, but I don't know what that'd be; pods should not be behind a service until they report healthy.

| 2025-02-13 13:55:58 MySQL_Monitor.cpp:4502:monitor_dns_resolver_thread(): [ERROR] An error occurred while resolving hostname:  [-2]                                      │
│ 2025-02-13 13:56:00 ProxySQL_Cluster.cpp:273:ProxySQL_Cluster_Monitor_thread(): [WARNING] Cluster: unable to connect to peer :6032 . Error: Can't connect to local serve │
│ 2025-02-13 13:56:02 ProxySQL_Cluster.cpp:273:ProxySQL_Cluster_Monitor_thread(): [WARNING] Cluster: unable to connect to peer :6032 . Error: Can't connect to local serve │
│ 2025-02-13 13:56:04 ProxySQL_Cluster.cpp:273:ProxySQL_Cluster_Monitor_thread(): [WARNING] Cluster: unable to connect to peer :6032 . Error: Can't connect to local serve │
│ 2025-02-13 13:56:06 ProxySQL_Cluster.cpp:273:ProxySQL_Cluster_Monitor_thread(): [WARNING] Cluster: unable to connect to peer :6032 . Error: Can't connect to local serve │
│ 2025-02-13 13:56:08 ProxySQL_Cluster.cpp:273:ProxySQL_Cluster_Monitor_thread(): [WARNING] Cluster: unable to connect to peer :6032 . Error: Can't connect to local serve │
│ 2025-02-13 13:56:08 [INFO] Received LOAD PROXYSQL SERVERS TO RUNTIME command