In my current job, I have the need to introduce container orchestration into our backend. While we have the classic VM orchestration system in place, it is time to move our smattering of standalone docker (moby?) hosts into a more coherent whole. Thus begins my journey into container orchestration starting with Kubernetes.
In this series of posts, I will walk through various well-known tutorials available. I hope to bring the “CCIE lab workbook” pedagogical method with me: that is extensive transcripts, where one could cut-and-paste the commands, and view detailed annotated console output. With this method, I hope to smooth the speed bumps that you and my team may encounter, as we bring up and tear down the labs, and try to reproduce the results.
If you have ever tried to learn a system with many moving parts such as Ceph or OpenStack and follow along the tutorials you will frequently encounter hair-pulling and caffeine-drowning moments: your test systems are just not behaving as the tutorial would suggest. Dedicated students of the various CCIE certifications will know what I mean — and they have the benefit of comprehensive workbooks with extensive logging recorded. With the excellent albeit terse tutorials available online, they frequently don’t have time or space to go into the various quirks that you may encounter.
Think of this as my virtual
CCIE Kubernetes workbook in the sky: for each tutorial, I will faithfully transcribe the command lines and output and report on the various issues that I encountered. In the interests of brevity, the full transcripts (command lines, console output, configuration files) will be available in a github repo. In each blog post, I will highlight the goals, the TL;DR, and the speed bumps along the way. There will also be posts where I tear down some subsystems and explore the way the tutorial set things up, and how it might differ from a production version.
There is a classic CCIE workbook where the playbook begins by going to eBay and picking up 6 routers (3 of which need to be x8xx series for MPLS), 4 switches (2 of which need to be 356x for whatever reason) and 3 backbone routers, not to mention a crazy array of synchronous serial ports — thankfully those days are gone. In the new world go to your VirtualBox-bay or KVM-bay and pick up 4 VMs and journey along with me. As we progress, I would love to get your feedback on whether you find this useful, and how the transcripts could be improved.
Many thanks to the PacketPushers team for allowing me to post here…and stay tuned for the first chapter in our workbook where we will set up a mini-cluster on four VMs with kubeadm.