witty-wall-71317
02/22/2025, 5:04 AMFailed to pull image "gardendev/k8s-util:0.6.2@sha256:f51e7ce040e2e23bc0eaa7216e4d976f13786d96773ef7b8c8f349e7a63d74e9"
The problem seems to be related to rate limits when pulling the gardendev/k8s-util image.
This is now happening more since spot instance is trying to place the pods in appropriate nodes more often. I’m looking for a way to specify a custom image hosted in my ECR repository to avoid hitting these rate limits.
I’m using Kaniko as the builder, and it already uses its own custom image, which is also hosted on ECR. Does anyone have any suggestions on how I can configure this to resolve the issue and prevent hitting these rate limits?
My current provider configuration looks something like this:
providers:
- name: kubernetes
context: arn:aws:eks:us-east-1:123456789:cluster/garden
buildMode: kaniko
kaniko:
image: 123456789.dkr.ecr.us-east-1.amazonaws.com/local-mirror:gcr.io-kaniko-project-executor-v1.11.0-debug
extraFlags:
- --snapshot-mode=redo
- --use-new-run
nodeSelector:
node/namespace: "build"
tolerations:
- key: node
value: build
namespace: dara-${local.username}
deploymentRegistry:
hostname: 123456789.dkr.ecr.us-west-2.amazonaws.com
namespace: garden
Also on a side note, I tried following the documentation here for utilImageRegistry setup https://docs.garden.io/reference/providers/kubernetes#complete-yaml-schema but it wasn't helpful as I kept getting error
Error validating provider configuration: key "utilImageRegistryDomain" is not allowed at path [utilImageRegistryDomain]. Available keys: dependencies, environments, buildMode, etc
I am using garden version: 0.13.13
just fyifreezing-pharmacist-34446
02/25/2025, 1:11 PMutilImageRegistryDomain
is the correct field for that. This field was only introduced in the 0.13.43 release see https://github.com/garden-io/garden/releases/tag/0.13.43 . Please upgrade garden to use this flag.