Startup Taint is sometimes not removed #70
Description
We're seeing an issue with the fsx driver where it doesn't remove the taint off a node sometimes, and has the error: "the server rejected our request due to an error in our request" on the node pod. Killing the pod makes the problem go away. We're using cilium, karpenter, and node startup taints for those (and fsx).
It's basically almost a word-for-word copy of this issue that the efs driver had, and i wonder if it's a similar solution needed ?: kubernetes-sigs/aws-efs-csi-driver#1320