blakethepatton
commented
2 years ago I would add that for chapter 2 I encountered issues running the playbook, specifically:
TASK [Create a Deployment for Hello Go.] *** fatal: [127.0.0.1]: FAILED! => {"changed": false, "msg": "Failed to get client due to HTTPConnectionPool(host='localhost', port=80): Max retries exceeded with url: /version (Caused by NewConnectionError('<urllib3.connection.HTTPConnection object at 0x7f4f9c9daf40>: Failed to establish a new connection: [Errno 111] Connection refused'))"}
Tracked the issue down to my version of ansible installed, described here: https://github.com/kubernetes-client/python/issues/1333
Solution was to update my install of ansible:
sudo apt unstall ansible
sudo add-apt-repository --yes --update ppa:ansible/ansible
sudo apt update
sudo apt install ansibleAfter updating, things worked.
Posting issues, notes & suggestions as I'm working along through the book:
My system:
Chapter 1
On Ubuntu,
minikube startnow uses "docker" as the default driver, and not "virtualbox" which you imply in the sections afterwards. To use "virtualbox" as the driver, one would need to runminikube start --driver=virtualbox. On subsequentminikube starts, the selected driver is saved in~/.minikubeand doesn't need to be specified anymore.Chapter 2
Section: Installing Ansible
pipandpython-devare not longer available through the default repo's. The default python version for Ubuntu/Debian is currently v3.8 and thus the commands need to be replaced bypip3andpython3-devChapter 3
Writing a Playbook to Build a Container Image
vars/main.ymlbut then use version "8.3.1" later on in "Writing a Playbook to Test the Container Image" where you usedocker run -d -p 8983:8983 ansible-for-kubernetes/solr:8.3.1to start the containerChapter 4
A Vagrantfile for local Infrastructure-as-Code
Running the cluster build playbook
ansible-playbook -i inventory main.ymlfails atTASK [geerlingguy.docker : Ensure dependencies are installed.]with the error:for each host (kube1, kube2, kube3). The error is probably related to the apt cache age of the used Vagrant boxed because I was able to fix it by running
sudo apt updateon each box (or Ansible-style:ansible -m command -a 'sudo apt update' -i inventory all. Running the playbook again then succeeds. An even better solution is to add the following to thepre_tasksof the playbook:Testing the cluster with a deployment using Ansible
TASK [Create hello-k8s resources and wait until they are Ready.]fails with errorFixed this issue by downgrading/pinning the openshift version in the
test-deployment.ymlplaybook to v0.11.2 as outlined in this issue.Patching Flannel to use the right network interface
diff -u kube-flannel.yml kube-flannel-virtualbox.yml > kube-flannel-patch.txt)kube-flannel-ds-amd64butkube-flannel-dsChapter 5
Authenticating to the EKS Cluster via kubeconfig
The
aws-iam-authenticatoris no longer required when usingaws-cliversion 1.16.156 or laterChapter 6
(TODO in book)
Chapter 7
Manage Kind with Molecule
The default YAML files generated with
molecule init scenarioseem to have changed (quite a bit). Following along and making changes as suggested result inmolecule testthrowing an error:I suspect this has to do with the absent
connection: localinconverge.ymlbecause after I modifiedconverge.ymlas outlined in the next subchapter of the book ("Test a playbook in Kind with Molecule"), the error went away.Test a playbook in Kind with Molecule
molecule convergesuccessfully runs (docker psshows me a running containerkindest/node:v1.20.2) but if I runkubectl get job helloafter that, I get an error:Seems like the new kubeconfig (
~/.kube/config-molecule-test) is not being used. Solved that by runningexport KUBECONFIG=~/.kube/config-molecule-testKubernetes CI Testing in GitHub Actions
molecule-kind.ymland later in the subchapter you say "Once you have theci.ymlworkflow file added to your repository"Groetjes, Ivo