Troubleshooting Kubernetes workload migrations to Amazon EKS

kubectl cannot connect to the source cluster

Verify the kubeconfig context is correct: kubectl config current-context. Check network connectivity and authentication credentials.

Extraction script fails with permission errors

Ensure the kubectl service account has cluster-admin or equivalent read permissions on the source cluster.

Migration dry-run shows CRD errors

Install the required CRD operators on the target Amazon EKS cluster before running the migration. The migration script does not install operators.

Workloads stuck in Pending state after migration

Check node capacity on the target Amazon EKS cluster (kubectl describe nodes). Verify that resource requests can be satisfied. Check for missing StorageClasses or PVCs.

Services not getting external IPs

Verify that the AWS Load Balancer Controller is installed and running. Check IAM permissions for the controller's service account.

ImagePullBackOff errors on migrated pods

Ensure Amazon EKS nodes have access to the container registry used by the source cluster. Configure Amazon ECR pull-through cache or image pull secrets as needed.

StorageClass provisioner mismatch

The migration script automatically remaps kubernetes.io/aws-ebs to ebs.csi.aws.com. For other provisioners, manually update the StorageClass before migration or after extraction.

Rollback script cannot find migration report

Specify the report path explicitly: python3 rollback_eks_migration.py --report <path-to-migration_report.json>.