multi-cluster support when using a kubeconfig
Darwin014 opened this issue · 4 comments
Hi!
Can I use aws-iam-authenticator for kubeconfig?
I installed aws-iam-authenticator in docker image and run pod but it doesn't seem to load kubeconfig.
[2021-10-20 09:08:31 +0000] [1] [INFO] Starting gunicorn 20.1.0
[2021-10-20 09:08:31 +0000] [1] [INFO] Listening at: http://0.0.0.0:8080 (1)
[2021-10-20 09:08:31 +0000] [1] [INFO] Using worker: gthread
[2021-10-20 09:08:31 +0000] [7] [INFO] Booting worker with pid: 7
[2021-10-20 09:08:31 +0000] [8] [INFO] Booting worker with pid: 8
[2021-10-20 09:08:41,849] INFO: RUNNING WITH AUTHENTICATION DISABLED
[2021-10-20 09:08:41,850] INFO: Attempting init with KUBECONFIG
[2021-10-20 09:08:41,859] INFO: KUBECONFIG loading failed but KUBERNETES_SERVICE_HOST environment variable found, assuming to be running inside a Kubernetes cluster
[2021-10-20 09:08:41,859] INFO: In cluster configuration loaded successfully.
[2021-10-20 09:08:42,048] INFO: RUNNING WITH AUTHENTICATION DISABLED
[2021-10-20 09:08:42,048] INFO: Attempting init with KUBECONFIG
[2021-10-20 09:08:42,056] INFO: KUBECONFIG loading failed but KUBERNETES_SERVICE_HOST environment variable found, assuming to be running inside a Kubernetes cluster
[2021-10-20 09:08:42,056] INFO: In cluster configuration loaded successfully.
Thanks
https://docs.aws.amazon.com/eks/latest/userguide/install-aws-iam-authenticator.html
https://docs.aws.amazon.com/eks/latest/userguide/create-kubeconfig.html
Hi @Darwin014
If I'm understanding the docs correctly, this should be transparent to GPM, the aws-iam-authenticator should update the kubeconfig with the right credentials.
Have you edited the docker image to run the command aws eks --region <region-code> update-kubeconfig --name <cluster_name> before GPM starts?
Could you tell us exactly what you have tried? could you also please test setting GPM in debug mode to get more details on what is failing?
Thanks.
Hi @Darwin014
We've seen some issues in AWS with v0.5.0 default memory limits that resulted in the behaviour you were seeing. Could you please try increasing them or even better switching to v0.5.1 and let us know if that works for you?
Thank you