destroy-hc-aws - SKILL.md Agent Skill

name: Destroy HC AWS description: "Destroy a HyperShift HostedCluster and all associated AWS infrastructure (VPC, IAM, Route53, etc.)."

Destroy HostedCluster

This skill destroys a HyperShift HostedCluster and all associated AWS infrastructure.

When to Use This Skill

Use this skill when:

You need to clean up a test HostedCluster
You want to destroy a HostedCluster and its AWS resources (VPC, subnets, NAT gateways, IAM roles, etc.)
You need to remove orphaned HostedClusters from previous test runs

Prerequisites

Source the environment file before using this skill:

source dev/claude-env.sh

Additional requirements:

AWS credentials loaded (source $AWS_CREDS_SOURCE)
KUBECONFIG pointing to management cluster ($MGMT_KUBECONFIG)
hypershift binary built (./bin/hypershift)

Environment Configuration

Environment variables from dev/claude-env.sh:

Variable	Description
`AWS_CREDENTIALS`	Path to AWS credentials file
`AWS_CREDS_SOURCE`	Script to source AWS env vars
`AWS_REGION`	AWS region
`MGMT_KUBECONFIG`	Path to management cluster kubeconfig

Command

source $AWS_CREDS_SOURCE && \
KUBECONFIG=$MGMT_KUBECONFIG \
./bin/hypershift destroy cluster aws \
  --name <CLUSTER_NAME> \
  --namespace <NAMESPACE> \
  --aws-creds $AWS_CREDENTIALS \
  --region $AWS_REGION

Parameters

Parameter	Description	Default
`--name`	Name of the HostedCluster to destroy	Required
`--namespace`	Namespace where the HostedCluster exists	`clusters`
`--aws-creds`	Path to AWS credentials file	`$AWS_CREDENTIALS`
`--region`	AWS region	`$AWS_REGION`

What Gets Destroyed

The command destroys:

HostedCluster and NodePool resources
HostedControlPlane namespace and all resources
AWS VPC and subnets
NAT gateways and internet gateways
Route tables
Security groups
DHCP options
Private hosted zones (Route53)
OIDC provider
IAM roles and instance profiles
VPC endpoints
Elastic IPs

Quick Cleanup for Orphaned HCs

If a HostedCluster is stuck deleting, you can force remove it:

# Remove cleanup annotation to skip cloud resource cleanup
KUBECONFIG=$MGMT_KUBECONFIG kubectl annotate hostedcluster <NAME> -n <NAMESPACE> \
  hypershift.openshift.io/cleanup-cloud-resources-

Example Usage

# List existing HostedClusters
KUBECONFIG=$MGMT_KUBECONFIG kubectl get hostedclusters -A

# Destroy a specific HostedCluster
source $AWS_CREDS_SOURCE && \
KUBECONFIG=$MGMT_KUBECONFIG \
./bin/hypershift destroy cluster aws \
  --name my-test-cluster \
  --namespace clusters \
  --aws-creds $AWS_CREDENTIALS \
  --region $AWS_REGION

Troubleshooting

NAT Gateway Still Deleting

NAT gateways take time to delete. The command will retry automatically.

VPC Has Dependencies

The command will retry until all dependencies (subnets, gateways, etc.) are deleted.

HostedCluster Stuck in Deleting

Check if there are stuck finalizers
Remove the cleanup-cloud-resources annotation if you want to skip AWS cleanup
Manually delete AWS resources if needed