puzzle / ansible-rancher Goto Github PK

Hi There

Sorry to bother you again. In the official Docker repo, used by the Docker role, the newest Docker version is 20.10.1. Which causes an unsupported Docker version error later in rke.

I worked around with fixing the version in group_vars:

# grep docker inventories/group_vars/all.yml
docker_package: docker-ce-19.03.13

That's not a perfect solution, but I'm not sure what a good solution would be. I see the following possibilities:

• Adding a version variable (One could think of a variable to set on 19.03 which then uses yum to search for the newest 19.03 release
• Make use of the Docker script of Rancherlabs (curl https://releases.rancher.com/install-docker/19.03.sh | sh) which is again not a perfect solution as it would run every time

What do you think?

BR
Thomas

https://rancher.com/docs/rancher/v2.x/en/installation/resources/advanced/firewall/

We recommend disabling firewalld. For Kubernetes 1.19, firewalld must be turned off.

Would you be interested in a PR if it's only tested against RHEL 7?

I've a question regarding the k8s roles assigned based on the inventory groups.

Given this sample inventory excerpt:

[mycluster_master]
node01
node02
node03

[mycluster_worker]
node01
node02
node03

All the nodes only get the worker role assigned. It completely ignores that they should (?) inherit the other two roles from the mycluster_master group. I solved this by adding all three roles to the mycluster_worker group.

I'm fully aware that this is

a) not the best idea of building a Kubernetes cluster
b) not supported by Rancher

but it's useful for testing and/or development scenarios. My question is: is this a issue of how this roles here are working? Or is it the behavior of Ansible? I assum the later, but I'm wondering.

When I use keepalived_setup_env=local, keepalive deployment fails:

TASK [rancher_keepalived : Create IP Failover DaemonSets for Internal IP] *********************************************************************************************************************************************
fatal: [rancher_mycluster]: FAILED! =>
  msg: |-
    The task includes an option with an undefined variable. The error was: 'cloudscale_api_token' is undefined

    The error appears to be in '/path/git/ansible-rancher/roles/rancher_keepalived/tasks/configure-keepalived.yml': line 105, column 3, but may
    be elsewhere in the file depending on the exact syntax problem.

    The offending line appears to be:


    - name: Create IP Failover DaemonSets for Internal IP
      ^ here

As soon as i set the token to a random string (eg. cloudscale_api_token="n/a") it works. It does not use the string at all, but it requires it.

Hi there

Environment:

ansible 2.9.15
python 3.6.8
CentOS Stream release 8

The roles/rke_rancher_clusters/tasks/rancher.yml makes use of the k8s module. Although it should exist in Ansible 2.9, I had no such module available on my machine. In Ansible 2.10, that particular module was moved to community. I installed it with ansible-galaxy collection install community.kubernetes and changed the k8s in roles/rke_rancher_clusters/tasks/rancher.yml to community.kubernetes.k8s. However, community.kubernetes.k8s has a dependency on the oc module. Which was deprecated and got removed in 2.9. They say, one should use openshift_raw instead, which is eventually the same as community.kubernetes.k8s.

Target Servers are all CentOS7

How did you work around that?

Kind regards
Thomas