You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When trying to deploy kubectl resources (such as KubernetesResource or HelmChart) against an EKS cluster with Fargate Profiles that are still being created, the API server may not be reachable. This causes these resources to to fail, the deployment to roll back and worst of all, as the rollback happens, CFN tries to delete the FargateProfile resources, but they still being creating. Basically leading to unrecoverable situation.
Environment
CLI Version :
Framework Version: 1.48.0
Node.js Version:
OS :
Language (Version):
This is 🐛 Bug Report
The text was updated successfully, but these errors were encountered:
When a Fargate profile is being created, the Kubernetes API server in EKS sometimes rejects requests. This means that kubectl-related resources such as KubernetesResources Helm charts may fail during deployment.
To address this, we add a "barrier resource" (in the form of an SSM parameter) which waits for all fargate profiles to be created before allowing kubectl resources to continue. This is done by the barrier taking a dependency on all FargateProfile resources and all kubectl resources taking a dependency on the barrier.
Fixes#8854
This commit also fixes#8574 by adding `iam:ListAttachedRolePolicies` to the cluster's creation role IAM policy.
When a Fargate profile is being created, the Kubernetes API server in EKS sometimes rejects requests. This means that kubectl-related resources such as KubernetesResources Helm charts may fail during deployment.
To address this, we add a "barrier resource" (in the form of an SSM parameter) which waits for all fargate profiles to be created before allowing kubectl resources to continue. This is done by the barrier taking a dependency on all FargateProfile resources and all kubectl resources taking a dependency on the barrier.
Fixes#8854
This commit also fixes#8574 by adding `iam:ListAttachedRolePolicies` to the cluster's creation role IAM policy.
…8859)
When a Fargate profile is being created, the Kubernetes API server in EKS sometimes rejects requests. This means that kubectl-related resources such as KubernetesResources Helm charts may fail during deployment.
To address this, we add a "barrier resource" (in the form of an SSM parameter) which waits for all fargate profiles to be created before allowing kubectl resources to continue. This is done by the barrier taking a dependency on all FargateProfile resources and all kubectl resources taking a dependency on the barrier.
Fixes#8854
This commit also fixes#8574 by adding `iam:ListAttachedRolePolicies` to the cluster's creation role IAM policy.
----
*By submitting this pull request, I confirm that my contribution is made under the terms of the Apache-2.0 license*
iliapolo
changed the title
[eks] Kubernetes resources fail to create if fargate profiles are not stable
[aws-eks] Kubernetes resources fail to create if fargate profiles are not stable
Aug 16, 2020
When trying to deploy
kubectl
resources (such asKubernetesResource
orHelmChart
) against an EKS cluster with Fargate Profiles that are still being created, the API server may not be reachable. This causes these resources to to fail, the deployment to roll back and worst of all, as the rollback happens, CFN tries to delete the FargateProfile resources, but they still being creating. Basically leading to unrecoverable situation.Environment
This is 🐛 Bug Report
The text was updated successfully, but these errors were encountered: