I'm moving my container workloads from AWS ECS to EKS and I'm wondering if Kubernetes has a similar circuit breaker feature. ECS has a mechanism that stops deployments after a certain number of failures occur, but my last experience with it was pretty frustrating because it didn't handle internal container errors well. With Kubernetes, I've noticed it just tries to launch pods but often ends up in CrashLoopBackoff. What options are available in Kubernetes to prevent this kind of situation?
1 Answer
The solution really depends on what tools you're using for deployment. If you're utilizing ArgoCD, it supports rollouts which can mimic the behavior of ECS's circuit breaker. On the other hand, if you’re using Flux, you might want to look into Helm releases. What does your deployment setup look like?
We use Terraform for our deployments. It's not the usual way, but we’re comfortable with HCL under pressure. We're planning to test out ArgoCD or Flux soon, so I’ll look into that!