This is my working directory for completing the CNCF Kubestronaut certification package.
This is not a pretty portfolio, it's getting things done dirty.
sources and inspirations include, but are not limited to:
- https://github.com/talbotfoundry/k8s-kvm
- https://github.com/boxcutter/kvm/
- https://github.com/chef/bento
When snippets are pulled from repo issues, they are commented inline with references.
Branches are squashed, because save scumming is not a sin.
created and ran on x86_64 Debian Linux.
virtual machines are x86_64 Ubuntu 24.04.1 Linux.
Vagrant is configured to copy ~/.ssh/id_rsa.pub to the virtual machine's root user's authorized keys. The vagrant makefile target will drop ssh configuration snippets into ~/.ssh/vagrant.d/.
Host vagrant-*
User root
RequestTTY yes
IdentityFile ~/.ssh/id_rsa.pub
Include vagrant.d/*
$ ssh vagrant-control-01 -- kubectl apply -f - < manifests/pod.yaml
pod/box created
$ scp vagrant-control-01:.kube/config ~/.kube/
$ kubectl get nodes
NAME STATUS ROLES AGE VERSION
vagrant-control-01 Ready control-plane 67m v1.32.1
vagrant-control-02 Ready control-plane 67m v1.32.1
[...]
installs local python / pyenv runtime. ansible, etc.
builds base images and imports to vagrant.
must be explicitly ran first. not included in default build targets / dependencies, as it is quite a heavy operation.
ubuntu24.pkr.hcl
: packer configuration file
scripts/packer.sh
: base box provision script
creates a 1 control node, 3 worker node k8s cluster by default.
export VAGRANT_K8S_PROFILE='full'
will create a 3 control node, 5 worker node cluster.
- killer.sh exam simulators included are invaluable for finding weak spots
- foo?
- Time and resource management.
- spend time wisely. if a question is taking a lot of digging through docs, flag and come back to it.
- Know key search terms for boilerplate in k8s official docs.
- vimrc
- add
set si
for smartindent
- add
- env
export d="--dry-run=client -o yaml"
- The CNCF Landscape is very intimidating.
- take time to become familiar with graduated projects.
-
cluster upgrade apply takes a very long time.
- tmux and detach. flag the task, and come back after a few minutes.
-
Time management:
- 30% troubleshooting
- 25% arch, install, configuration
- both are quick work if you know where to look.
- considerable overlap from CKA.
- time management - topics are more even in weighting.
- another large subject overlap with CKA, but multiple choice!
- Required a pretty massive shift in local vagrant configuration.
- Stuff that realistically should have been done for CKA.
- Admin users and etcd encryption forced it.