Stagger start RKE2 server nodes about rke2-ansible HOT 7 CLOSED

Hi,
I think there is another possible solution. An another point of view.
If I understand correctly, the idea is wait until the kubeapi is up and running ?

There is a more "ansible" approach with the module: ansible.builtin.wait_for
You can add :

- name: Wait for k8s apiserver
  wait_for:
    host: "{{ kubernetes_api_server_host }}"
    port: "6443"
    state: present
    timeout: 300

This is what kubespray seems to do that here:
https://github.com/kubernetes-sigs/kubespray/blob/33a60fe919b50a149ef6402f56e342989f6eea23/roles/kubernetes/client/tasks/main.yml#L44

And if you just want to verify if each node are up and running, I think kubectl command is a good approach, after checking that the kube api is up.
This approach, if you create an ansible role can also be useful for issue #15 : rolling upgrade playbook and node addition playbook.

What do you think ?

from rke2-ansible.

This seems related to #38.

I moved past this frustrating issue using @AYDEV-FR 's solution at the bottom of the rke2_server > first_server.yaml task, along with adding serial: 1 to the site.yml rke2_servers playbook. Thanks for that.

However, I think a watch script that isn't tied to a play might have more flexibility for re-use.

from rke2-ansible.

should be addressed in #54

from rke2-ansible.

@mddamato Yes, the introduction of the throttle directive here solved the problems we had with the 3rd server hanging on startup in a 3-server cluster deployment. Thanks.

from rke2-ansible.

You might wait until the necessary commit hits main before closing this issue.

from rke2-ansible.

I'm having the same issue right now. Why is this issue closed?

TASK [rke2_server : Start rke2-server] ***************************************** ok: [rkectlplane01] Wednesday 28 September 2022 00:46:09 +0000 (0:00:01.183) 0:00:36.379 *** TASK [rke2_server : Wait for k8s apiserver] ************************************ fatal: [rkectlplane01]: FAILED! => {"changed": false, "elapsed": 300, "msg": "Timeout when waiting for rkectlplane01:6443"} NO MORE HOSTS LEFT *************************************************************

from rke2-ansible.

@siulcode The error you posted shows that the server isn't starting properly but it's not enough information for me to go off of. Could you post the errors from the rke2 server logs or other applicable components?

from rke2-ansible.