I created a EKS cluster with Localstack, which starts fine. I then target my awslocal to the cluster as shown in the documentation. Afterward I try to install an ingress controller via a helm chart into the cluster. The pod which is created can not pull the image from the public docker registry.
Logs from EKS POD:
2023-12-06 09:58:09 E1206 14:58:09.994663 23 pod_workers.go:918] "Error syncing pod, skipping" err="failed to \"StartContainer\" for \"create\" with ImagePullBackOff: \"Back-off pulling image \\\"registry.k8s.io/ingress-nginx/kube-webhook-certgen:v20231011-8b53cabe0@sha256:a7943503b45d552785aa3b5e457f169a5661fb94d82b8a3373bcd9ebaf9aac80\\\"\"" pod="ingress-nginx/ingress-nginx-admission-create--1-6npf6" podUID=4ebbb395-a104-4a53-bd92-f53ba5eb7675
Looking for any help or ideas on how to address this problem. I would prefer not to have to re-tag everything in this helm chart to a localstack ECR registry.
Hi @CharminUltra561, thanks for raising this issue. Would you mind restarting LocalStack with DEBUG=1, re-running your helm chart and pasting the verbose logs from the LocalStack container here? If you do not wish to share your logs here, you can open a ticket with our support bot on our community slack channel and one of our support engineers will be in touch.
Hi Simon I opened a ticket on the slack community bot and am awaiting response
A quick update: if you start LocalStack with the variable DNS_LOCAL_NAME_PATTERNS=prod-registry-k8s-io-.*\.s3\.dualstack\..*\.amazonaws\.com you can pull public docker images from registry.k8s.io. We need to update the documentation to reflect this.
A quick note: we will soon be deprecating DNS_LOCAL_NAME_PATTERNS, and will expect users to use DNS_NAME_PATTERNS_TO_RESOLVE_UPSTREAM instead. This change was merged today in our open source repository, however it has not made it in to the localstack/localstack-pro docker image. Just a heads up that you will see deprecation warnings in the future.