During a recent admin cluster upgrade from Anthos on-prem v1.15 to v1.16 we encountered this issue regarding duplicate hostnames in the same datacentre. In an attempt to resolve this we updated the hostnames in the admin-cluster-ipblock file and force updated the cluster. This updated the names of two of the nodes but failed to change the name of the master node.
This is the new ipblock file:
blocks:
- netmask: "255.255.255.192"
gateway: "10.0.99.60"
ips:
- ip: "10.0.99.15"
hostname: "admin-non-prod-1"
- ip: "10.0.99.16"
hostname: "admin-non-prod-2"
- ip: "10.0.99.17"
hostname: "admin-non-prod-3"
- ip: "10.0.99.18"
hostname: "admin-non-prod-4"
And these are our nodes following the update:
NAME STATUS ROLES AGE VERSION
admin-host-1 Ready control-plane,master 3h45m v1.27.4-gke.1600
admin-non-prod-2 Ready <none> 3h15m v1.27.4-gke.1600
admin-non-prod-4 Ready <none> 3h10m v1.27.4-gke.1600
The output of kubectl describe cluster shows that the Reserved Addresses has been updated as per the ipblock file:
Reserved Addresses:
Gateway: 10.0.99.60
Hostname: admin-non-prod-1
Ip: 10.0.99.15
Netmask: 26
Gateway: 10.0.99.60
Hostname: admin-non-prod-2
Ip: 10.0.99.16
Netmask: 26
Gateway: 10.0.99.60
Hostname: admin-non-prod-3
Ip: 10.0.99.17
Netmask: 26
Gateway: 10.0.99.60
Hostname: admin-non-prod-4
Ip: 10.0.99.18
Netmask: 26
How can we update the name of the master node?