[HarshCasper]

The repo had no CI, and setup.sh failed on current LocalStack.

Fixes to setup.sh:

• The default K8S_VERSION was 1.29, which LocalStack EKS now rejects with unsupported Kubernetes version 1.29 (it supports 1.30 to 1.36). The default is now 1.35, and the README lists the variable.
• The LocalStack wait looked for "eks": "running" with no limit. EKS reports available until its first request, so it now accepts either, and gives up after MAX_WAIT_SECONDS (default 600).
• The wait for the cluster to become ACTIVE has the same limit.
• k3d image import --cluster lifecycle-demo targeted a cluster that doesn't exist, because LocalStack appends a hash to the k3d cluster name (lifecycle-demo-1ac9fa4e in CI). It now looks the name up with k3d cluster list and falls back to the docker exec import if it finds none. The README troubleshooting step says the same.

New test.sh redeploys the app and checks each step the demo shows:

• the API server accepts the manifest
• the Deployment creates a ReplicaSet and 2 Ready pods
• each pod is assigned to a node, with Scheduled and Started events
• both pods are Service endpoints
• 30 requests through the Service reach both pods, and each reports its own pod and node name through the Downward API
• a deleted pod is replaced, and the replacement becomes Ready

New workflow installs lstk and k3d, runs ./setup.sh, AUTO=1 ./demo.sh and ./test.sh, and prints the cluster state and LocalStack logs on failure. The README documents ./test.sh.

Verified by CI: https://github.com/localstack-samples/k8s-lifecycle-demo/actions/runs/38407304457. The demo runs all 7 steps and test.sh prints all 6 PASS lines.

https://claude.ai/code/session_01DZfDmzUBPqNAu3yFKyfCNM