Failed to remove or reset node when number of etcd hosts is not divisible by 2 #12268
Closed as not planned
Description
What happened?
In a cluster with 7 etcd hosts, I removed an etcd host using remove_node.yml playbook. Then when I tried to remove another etcd host using remove_node.yml playbook, it failed at:
TASK [validate_inventory : Stop if even number of etcd hosts] ******************
fatal: [test-rm-c1]: FAILED! => {
"assertion": "groups.get('etcd', groups.kube_control_plane) | length is not divisibleby 2",
"changed": false,
"evaluated_to": false,
"msg": "Assertion failed"
}
Running reset.yml also failed at the same place.
Issue is introduced by commit 1b99195 where validate_inventory role is added to boilerplate.yml playbook.
What did you expect to happen?
I can remove a node from a cluster with even number of etcd hosts (e.g. to make the cluster with odd number of etcd hosts).
I should also be able to reset a cluster with even number of etcd hosts.
How can we reproduce it (as minimally and precisely as possible)?
Run the remove_node.yml and reset.yml playbooks with even number of etcd hosts.
OS
RHEL 8
Version of Ansible
core 2.16.14
Version of Python
3.12.8
Version of Kubespray (commit)
v2.28.0
Network plugin used
flannel
Full inventory with variables
Command used to invoke ansible
Output of ansible run
Anything else we need to know
No response