Two new test blocks in the existing k8s-deploy run-deploy-test.sh:
1. After first --perform-cluster-management start, assert that
caddy-cert-backup SA/Role/RoleBinding/CronJob were deployed by
install_ingress_for_kind.
2. Before final teardown, a full backup→restore cycle: seed a fake
manager=caddy Secret, trigger the CronJob immediately (kubectl
create job --from=cronjob), verify the backup file appeared under
{kind-mount-root}/caddy-cert-backup/caddy-secrets.yaml with the
seeded value, then stop --perform-cluster-management (destroys
Kind cluster), start --perform-cluster-management (fresh cluster,
runs restore before Caddy Deployment), assert the seeded Secret
was recreated with matching data.
kind-mount-root points at ~/stack-orchestrator-test/kind-mount
(sibling of CERC_REPO_BASE_DIR, outside any deployment dir).
Previous runs' root-owned files are cleaned via sudo rm -rf at start.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
c835ad7fdf