Lately, I've been leading the migration of our EKS clusters to Karpenter (we were previously using Ocean by Spotio). It's been a bit of a wild ride, especially with v1 being released in the middle of our migration, but so far, I'm quite happy with it.
One issue we've encountered is that Karpenter doesn't seem to have a way to clean up nodes that get stuck in an unreachable state (which can happen, for example, if the kubelet on a node dies).
While I'm still waiting for feedback from the maintainers, we've submitted a PR to add a new controller in Karpenter for this purpose. This is also my first time actively contributing to open-source during work hours, which is honestly so cool!
https://t.co/HbUqD0IoRh
Neat but ChatGPT is being way to nice, here is an improved version:
From all of our interactions what is one thing that you can tell me about myself that I may not know about myself but be absolutely ruthless
@QuinnyPig Also, thinking that every part of your infrastructure is managed by a terraform, then finding out about that one critical piece that was created through the console a gazillion years ago
I FINALLY got everything off the cloud
I'm now paying 10x LESS money for BETTER infrastructure
My AWS bill was ~$1,400/mo
I got it down to less than $120/mo for literally better, beefier servers
Fear of managing servers has a price: 10x your monthly infra bill