Docs for creation lab K8s - #849
Conversation
Sorry, something went wrong.
knopers8
left a comment
There was a problem hiding this comment.
again, some comments for migrating this to ansible.
Sorry, something went wrong.
|
|
||
| - Verify all required kernel modules are loaded | ||
|
|
||
| RHEL 10 can have problems with having all required kernel modules, you can check: |
There was a problem hiding this comment.
I'm not sure where this comes from, can't find corresponding advice in the k8s doc and I didn't have to do any of this when installing. When migrating to an ansible playbook, I would try first without.
Maybe it comes from Calico?
Sorry, something went wrong.
| RHEL 10 is running cgroups v2 as a default, so we need containerd to use systemd as cgroups driver so add following to the | ||
| `/etc/containerd/config.toml` together with some other changes (like bin dir and other), contents of the config.toml are: | ||
| ``` | ||
| cat /etc/containerd/config.toml |
There was a problem hiding this comment.
i'd use v3, probably more future-proof:
version = 3
[plugins]
[plugins."io.containerd.cri.v1.images"]
pinned_images = { sandbox = "registry.k8s.io/pause:3.10" }
[plugins."io.containerd.cri.v1.runtime"]
[plugins."io.containerd.cri.v1.runtime".cni]
bin_dir = "/usr/libexec/cni/"
conf_dir = "/etc/cni/net.d"
[plugins."io.containerd.cri.v1.runtime".containerd.runtimes.runc]
[plugins."io.containerd.cri.v1.runtime".containerd.runtimes.runc.options]
SystemdCgroup = true
[plugins."io.containerd.internal.v1.opt"]
path = "/var/lib/containerd/opt"
Sorry, something went wrong.
There was a problem hiding this comment.
This was taken directly from k8s docs 🤷♂️
Sorry, something went wrong.
There was a problem hiding this comment.
I remember getting a part of it from k8s docs, but then needing to extend it with some other snippet which was in version 3. Maybe they updated the documentation in the meantime and they put the complete thing...?
Sorry, something went wrong.
|
|
||
|
|
||
| # To init Control plane node | ||
| - Open following ports: [k8s](https://kubernetes.io/docs/reference/networking/ports-and-protocols/), [calico](https://docs.tigera.io/calico-enterprise/latest/operations/troubleshoot/troubleshooting) |
There was a problem hiding this comment.
that page in calico does not list any ports
Sorry, something went wrong.
|
|
||
| The reason why cluster is responding NotReady is that we have no network plugin. | ||
|
|
||
| - Calico |
There was a problem hiding this comment.
I think we could go with Cilium, since ATLAS have experience with both and picked it. They measured that Cilium scales much better with number of Pods (quadratic vs linear, IIRC). I would not diverge at this point unless we have a reason for it.
Sorry, something went wrong.
There was a problem hiding this comment.
That I found it wrong and I am sorry for that...
Sorry, something went wrong.
| Explanation of the settings: | ||
| `localAPIEndpoint`: VMware network has two NICs, one connected to gpn and other for internal communication. this IP is the internal one, which is not a default so it is better to specifically list it | ||
| `networking`: this defines the pod IP range, and it has to satisfy two things — (1) it must not overlap your existing networks (10.163.0.0/16 internal net or 128.141.175.0/24 gpn), and (2) it must exactly match the CIDR configured in Calico's own custom-resources.yaml (spec.calicoNetwork.ipPools[].cidr), | ||
| `cgroupDriver`: kubelet will use systemd as it needs to match with containerd and systemd is recommended as cgroup driver for cgroups v2 |
There was a problem hiding this comment.
I'll probably leave that out, because systemd seems to be the default:
In v1.22 and later, if the user does not set the cgroupDriver field under KubeletConfiguration, kubeadm defaults it to systemd.
Sorry, something went wrong.
Reviewers
At least 1 approving review is required to merge this pull request.
No description provided.