It's Okay To Be Wrong
Hands-down the hardest part of this field is the stress of things not working. When things break, you can freeze up, loose control, and sometimes just blank. Most of the time, the answer’s right in front of you, but something about the necessity to reach that highest mark keeps you on edge and the answer out of reach.
Sometimes I think it’s underestimated, the importance of admitting failure. Admitting that you didn’t know the answer - or were simply going about something the entirely wrong way - shows you are aware of your limitations. Being aware is half the battle to being prepared.
So here’s my little fuck-up for the evening.
For the last 2-3 weeks I’ve been working porting this website to Kube. Today, I started by laying out some simple terraform for a Linode kube cluster. When it came time to install resources onto Kube itself, I thought “If terraform offers a kube provider, why not build all my resources right alongside the infrastructure?”
Well, while I was in the process of playing with the kube provider, I ran across this article:
And to truly show the levels of insistent stupidity I can reach, just know that article didn’t stop me from me trying. Yes, that’s right. I read the contents of that article, which clearly link to the Terraform Kubernetes provider stating not to do this, and went about doing that anyways.
Sure enough, the first time I tore the infrastructure down and rebuilt it, low and behold:
Error: Kubernetes cluster unreachable: invalid configuration: no configuration has been provided, try setting KUBERNETES_MASTER environment variable
Get "http://localhost/api/v1/namespaces/static-site": dial tcp [::1]:80: connect: connection refused
A whole day put into writing some nice code only to find I’ll have to go about this the way everyone else told me to and use manifest files after all.
So what’s the take away, you may be wondering? What was my mistake?
I didn’t read. I didn’t listen. I didn’t accept my lack of knowledge around a subject and accept that others may have more to offer. And in this case, I really didn’t read.
Because of that, I failed. Tomorrow, when I read an article that clearly tells me the pitfalls of going a specific route, I’ll take the advice and save myself an evening.
Update 6/21/2022
Cloudflare had an outage this morning. I couldn’t help but think of this article while reading over the post-mortem. This was handled excellently and swiftly, especially considering the conditions they were working under after the mistake had been rolled out. It’s an excellent write up and worth the read.
Later today, I spotted this reddit post as well. Very Cool!