-
Notifications
You must be signed in to change notification settings - Fork 748
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
EKS + VPC CNI + Containerd works super slow #1509
Comments
Hi @ash-sdt Regarding the restart -
Can you please check if kube-proxy is failing to get node info? Ref - #1078 (comment) Based on the logs I see WARM_ENI_TARGET is set to 1 and the assigned IPs is 0 hence you will see just one ENI - |
@jayanthvn It seems it can't
|
You can try this workaround - kubernetes/kubernetes#61486 (comment) and this should fix kube-proxy error. We are also working internally to fix this up in kubeproxy. regarding the 2nd issue, can you please confirm if you are able to attach ENIs when you scale the pods? |
Thank you @jayanthvn |
Well it seems there is another problem now.
|
Well it seems the DNS issue is not related to VPC-CNI. |
What happened:
I am running an EKS cluster with self-managed Debian worker nodes and I use containerd cri instead of dockershim. I pass the containerd cri socket to "/var/run/cri.sock" in aws-node pods.
When I spin-up a new node, IPAMD acts super slow and makes the node unusable for a great amount of time. First it takes a lot of time to pass "Testing communication with server" stage, it fails the first time and passes in the second time. this causes aws-node pods to get restarded. also does not add ENI to my worker nodes. If I restart the nodes, it may add ENI to them after some time.
I have also tested Ubuntu Focal, same problem. However, it does not happen with official AL2 EKS AMI.
Unfortunately, there is nothing useful in logs.
These are logs from a test cluster:
Environment:
The text was updated successfully, but these errors were encountered: