feat(deploy): bootstrap deploy/.env and add auth-enabled flag (FLPATH-4806)

.cursor/prompts/deploy-dcm.md

@@ -48,6 +48,19 @@ Deploy the full DCM stack for E2E testing using `scripts/deploy-dcm.sh`.
 ./scripts/deploy-dcm.sh --cleanup-on-failure
 ```
 
+### Deploy with Authentication Enabled
+```bash
+# Starts Keycloak (compose profile auth) and enables JWT validation
+./scripts/deploy-dcm.sh --auth-enabled
+
+# Equivalent via environment (Jenkins uses this today)
+AUTH_DISABLED=false AUTH_ISSUER_URL=http://keycloak:8080/realms/dcm AUTH_JWT_AUDIENCE=dcm-api \
+    ./scripts/deploy-dcm.sh
+
+# Tear down auth-enabled stack with the same flag
+./scripts/deploy-dcm.sh --auth-enabled --tear-down
+```
+
 ### Deploy with k8s Container Service Provider
 ```bash
 # Auto-detects cluster from existing oc/kubectl session
@@ -107,17 +120,23 @@ When any service provider is enabled, the script resolves cluster access in this
 | `OPENSHIFT_API` | `--cluster-api` |
 | `OPENSHIFT_USERNAME` | `--cluster-username` |
 | `OPENSHIFT_PASSWORD` | `--cluster-password` |
+| `AUTH_DISABLED` | Set to `false` to enable auth (same as `--auth-enabled`) |
 
 Flags take precedence over environment variables.
 
+## Compose Credentials
+
+After cloning control-plane, the script creates `deploy/.env` from `deploy/.env.example` if missing. Database and optional auth credentials are written there for compose `env_file: .env` services. Lab defaults match control-plane's `.env.example`; override via shell env vars before running the script.
+
 ## What Happens
 
 1. Clones control-plane (`deploy/compose.yaml`)
-2. Runs `podman-compose up -d`
-3. Verifies all containers are running
-4. Polls `/api/v1alpha1/health` (90s timeout)
-5. Resolves container images to git commit SHAs via Quay.io API
-6. Writes `dcm-versions.json`
+2. Bootstraps `deploy/.env` with DB credentials (and auth credentials when `--auth-enabled`)
+3. Runs `podman-compose up -d`
+4. Verifies all containers are running
+5. Polls `/api/v1alpha1/health` (90s timeout)
+6. Resolves container images to git commit SHAs via Quay.io API
+7. Writes `dcm-versions.json`
 
 ## Output
 

.cursor/prompts/tear-down.md

@@ -14,6 +14,13 @@ Stop and clean up a running DCM deployment.
 ./scripts/deploy-dcm.sh --control-plane-dir /path/to/deploy --tear-down
 ```
 
+### Authenticated stack
+```bash
+./scripts/deploy-dcm.sh --auth-enabled --tear-down
+```
+
+Use the same authentication setting used during deployment so the auth compose profile, including Keycloak, is included in teardown.
+
 ## What Happens
 
 1. Stops all containers via `podman-compose down -v`

.github/workflows/validate-scripts.yaml

@@ -98,6 +98,7 @@ jobs:
             --expect-contains "--control-plane-branch" "control-plane-branch flag" \
             --expect-contains "--kubeconfig" "kubeconfig flag" \
             --expect-contains "--compose-file" "compose-file flag" \
+            --expect-contains "--auth-enabled" "auth-enabled flag" \
             --expect-contains "--cleanup-on-failure" "cleanup-on-failure flag" \
             --expect-contains "--running-versions" "running-versions flag" \
             --expect-contains "--tear-down" "tear-down flag" \

CLAUDE.md

@@ -28,7 +28,9 @@ CI runs ShellCheck on changed `*.sh` files via `.github/workflows/lint.yaml` (on
 
 Deploys the full DCM stack for E2E testing by cloning control-plane (`deploy/compose.yaml`), running `podman-compose up`, and polling health endpoints until all services respond 2xx.
 
-**Flow:** clone control-plane → `podman-compose up -d` → verify containers running → poll `/api/v1alpha1/health` → collect container versions from Quay.io API → write `dcm-versions.json`.
+**Flow:** clone control-plane → bootstrap `deploy/.env` → `podman-compose up -d` → verify containers running → poll `/api/v1alpha1/health` → collect container versions from Quay.io API → write `dcm-versions.json`.
+
+**Compose credentials:** After clone, the script copies `deploy/.env.example` to `deploy/.env` when missing and upserts DB/auth keys (lab defaults unless overridden by shell env). Control-plane compose reads these via `env_file: .env`. Pass `--auth-enabled` or set `AUTH_DISABLED=false` to add the compose `auth` profile (Keycloak) and write auth credentials into `.env`.
 
 **Modes:** The script has three mutually exclusive modes:
 - **Deploy** (default): full clone + bring-up + health check. Pass `--cleanup-on-failure` to auto-teardown on error (default leaves partial state for debugging).
@@ -48,6 +50,8 @@ When a non-main version is specified, `--control-plane-branch` is auto-derived t
 
 **Cluster authentication:** When any provider is enabled, the script resolves cluster access in priority order: explicit `--kubeconfig`, existing `oc`/`kubectl` session, or `oc login` via `--cluster-api` + `--cluster-password`.
 
+**Control-plane authentication:** Pass `--auth-enabled` (or set `AUTH_DISABLED=false`) to start Keycloak and enable JWT validation. Use the same flag on `--tear-down` when tearing down an auth-enabled stack. The E2E suite currently supports unauthenticated test runs only.
+
 Run `./scripts/deploy-dcm.sh --help` for all flags and environment variable overrides.
 
 ## Local dev scripts
@@ -107,6 +111,8 @@ The script is organized into sections separated by comment banners. Key function
 | `validate_k8s_container_provider` | Validates k8s container SP prerequisites |
 | `validate_k8s_storage_provider` | Validates k8s storage SP prerequisites |
 | `validate_acm_cluster_provider` | Validates ACM cluster SP prerequisites |
+| `ensure_deploy_env` | Bootstraps `deploy/.env` from `.env.example` and upserts credentials |
+| `upsert_deploy_env_var` | Idempotently sets a key in `deploy/.env` |
 | `resolve_provider_cli` | Resolves `oc`/`kubectl` per provider's `CLI_REQUIREMENT` |
 | `collect_provider_compose` | Collects compose profiles/overrides for an enabled provider |
 | `verify_health` | Confirms all compose services are running, then polls health endpoints with timeout |

README.md

@@ -39,9 +39,10 @@ Both deploy mode and `--running-versions` produce a `dcm-versions.json` mapping
 `scripts/deploy-dcm.sh` automates the full DCM stack lifecycle for E2E testing:
 
 1. Clones the [control-plane](https://github.com/dcm-project/control-plane) repo (`deploy/compose.yaml`)
-2. Starts all services with `podman-compose up`
-3. Polls health endpoints until every service responds 2xx
-4. Resolves running container images to git commit SHAs via the Quay.io API
+2. Bootstraps `deploy/.env` from `deploy/.env.example` (compose credentials; see control-plane `deploy/RUN.md`)
+3. Starts all services with `podman-compose up`
+4. Polls health endpoints until every service responds 2xx
+5. Resolves running container images to git commit SHAs via the Quay.io API
 
 ### Prerequisites
 
@@ -78,8 +79,11 @@ Both deploy mode and `--running-versions` produce a `dcm-versions.json` mapping
 # 8. Deploy ACM cluster provider (install ACM first if needed)
 ./scripts/deploy-dcm.sh --acm-cluster-service-provider --deploy-acm --kubeconfig ~/.kube/config
 
-# 9. Tear down when done
-./scripts/deploy-dcm.sh --tear-down
+# 9. Deploy with authentication enabled (Keycloak + JWT validation)
+./scripts/deploy-dcm.sh --auth-enabled
+
+# 10. Tear down when done
+./scripts/deploy-dcm.sh --auth-enabled --tear-down
 ```
 
 Run `./scripts/deploy-dcm.sh --help` for all flags and environment variable overrides.

scripts/deploy-dcm.sh

@@ -145,6 +145,7 @@ EOF
   --cluster-username USER        Username for oc login (default: kubeadmin)
   --cluster-password PASS        Password for oc login
   --compose-file PATH            Additional compose file to merge (repeatable, e.g. port overrides)
+  --auth-enabled                 Enable authentication (Keycloak + JWT validation; adds compose profile auth)
   --cleanup-on-failure           Tear down the stack automatically if deployment fails (default: leave for debugging)
   --running-versions             Print versions of all running containers and write dcm-versions.json
   --tear-down                    Stop the stack, remove volumes, and clean the deploy directory
@@ -165,6 +166,7 @@ Environment variables (flags take precedence):
   OPENSHIFT_API             Same as --cluster-api
   OPENSHIFT_USERNAME        Same as --cluster-username (default: kubeadmin)
   OPENSHIFT_PASSWORD        Same as --cluster-password
+  AUTH_DISABLED             Set to 'false' to enable auth (same effect as --auth-enabled)
 EOF
 
     # Provider namespace env vars (generated from registry)
@@ -189,6 +191,7 @@ Examples:
   $(basename "$0") --k8s-storage-service-provider --kubeconfig ~/.kube/config
   $(basename "$0") --all-service-providers --cluster-api https://api.cluster.example.com --cluster-password secret
   $(basename "$0") --acm-cluster-service-provider --deploy-acm --kubeconfig ~/.kube/config
+  $(basename "$0") --auth-enabled
   $(basename "$0") --tear-down
   $(basename "$0") --running-versions
 EOF
@@ -605,6 +608,115 @@ resolve_provider_cli() {
     esac
 }
 
+# --- Compose credential bootstrap ------------------------------------------ #
+#
+# control-plane deploy/compose.yaml reads credentials from deploy/.env
+# (see control-plane deploy/.env.example). Bootstrap that file after clone.
+
+env_or_default() {
+    local var_name="$1"
+    local default_value="$2"
+    if [[ -n "${!var_name:-}" ]]; then
+        echo "${!var_name}"
+    else
+        echo "${default_value}"
+    fi
+}
+
+resolve_db_password() {
+    local password_var
+    local resolved_password=""
+    local explicit_password=""
+
+    for password_var in POSTGRES_PASSWORD DB_PASS DB_PASSWORD; do
+        if [[ -n "${!password_var:-}" ]]; then
+            if [[ -z "${explicit_password}" ]]; then
+                explicit_password="${!password_var}"
+                resolved_password="${password_var}=${!password_var}"
+            elif [[ "${!password_var}" != "${explicit_password}" ]]; then
+                err "Conflicting database passwords supplied via ${resolved_password%%=*} and ${password_var}"
+                return 1
+            fi
+        fi
+    done
+
+    if [[ -z "${explicit_password}" ]]; then
+        explicit_password="adminpass"
+    fi
+
+    printf '%s\n' "${explicit_password}"
+}
+
+upsert_deploy_env_var() {
+    local deploy_dir="$1"
+    local key="$2"
+    local value="$3"
+    local env_file="${deploy_dir}/deploy/.env"
+    local tmp
+
+    [[ -n "${value}" ]] || return 0
+
+    tmp="$(mktemp)"
+    if [[ -f "${env_file}" ]]; then
+        grep -v "^${key}=" "${env_file}" > "${tmp}" || true
+    fi
+    printf '%s=%s\n' "${key}" "${value}" >> "${tmp}"
+    mv "${tmp}" "${env_file}"
+}
+
+ensure_deploy_env() {
+    local deploy_dir="$1"
+    local env_file="${deploy_dir}/deploy/.env"
+    local env_example="${deploy_dir}/deploy/.env.example"
+    local var
+    local db_password
+
+    if [[ ! -f "${env_file}" ]]; then
+        if [[ ! -f "${env_example}" ]]; then
+            err "Missing ${env_example} — cannot bootstrap deploy credentials"
+            return 1
+        fi
+        cp "${env_example}" "${env_file}"
+        info "Created ${env_file} from .env.example"
+    fi
+
+    db_password="$(resolve_db_password)" || return 1
+    upsert_deploy_env_var "${deploy_dir}" "POSTGRES_USER" "$(env_or_default POSTGRES_USER admin)"
+    upsert_deploy_env_var "${deploy_dir}" "POSTGRES_PASSWORD" "${db_password}"
+    upsert_deploy_env_var "${deploy_dir}" "DB_USER" "$(env_or_default DB_USER admin)"
+    upsert_deploy_env_var "${deploy_dir}" "DB_PASS" "${db_password}"
+    upsert_deploy_env_var "${deploy_dir}" "DB_PASSWORD" "${db_password}"
+
+    if [[ "${AUTH_ENABLED}" == true ]]; then
+        upsert_deploy_env_var "${deploy_dir}" "KEYCLOAK_ADMIN" "$(env_or_default KEYCLOAK_ADMIN admin)"
+        upsert_deploy_env_var "${deploy_dir}" "KEYCLOAK_ADMIN_PASSWORD" "$(env_or_default KEYCLOAK_ADMIN_PASSWORD admin)"
+        upsert_deploy_env_var "${deploy_dir}" "DCM_DEV_USER_PASSWORD" "$(env_or_default DCM_DEV_USER_PASSWORD admin)"
+        upsert_deploy_env_var "${deploy_dir}" "AUTH_PROXY_SECRET" "$(env_or_default AUTH_PROXY_SECRET dcm-dev-proxy-secret)"
+        upsert_deploy_env_var "${deploy_dir}" "AUTH_DISABLED" "false"
+        upsert_deploy_env_var "${deploy_dir}" "AUTH_ISSUER_URL" "$(env_or_default AUTH_ISSUER_URL http://keycloak:8080/realms/dcm)"
+        upsert_deploy_env_var "${deploy_dir}" "AUTH_JWT_AUDIENCE" "$(env_or_default AUTH_JWT_AUDIENCE dcm-api)"
+        upsert_deploy_env_var "${deploy_dir}" "DCM_ADMIN_SUBJECT" "$(env_or_default DCM_ADMIN_SUBJECT 56deb662-4820-5d83-b828-f4beb11a5fa7)"
+    else
+        upsert_deploy_env_var "${deploy_dir}" "AUTH_DISABLED" "true"
+    fi
+
+    if [[ -n "${DCM_VERSION:-}" ]]; then
+        for var in "${VERSION_ENV_VARS[@]}"; do
+            upsert_deploy_env_var "${deploy_dir}" "${var}" "${DCM_VERSION}"
+        done
+    else
+        for var in "${VERSION_ENV_VARS[@]}"; do
+            if [[ -n "${!var:-}" ]]; then
+                upsert_deploy_env_var "${deploy_dir}" "${var}" "${!var}"
+            fi
+        done
+    fi
+
+    if [[ -n "${ACM_CLUSTER_SP_PULL_SECRET:-}" ]]; then
+        upsert_deploy_env_var "${deploy_dir}" "ACM_CLUSTER_SP_PULL_SECRET" "${ACM_CLUSTER_SP_PULL_SECRET}"
+    fi
+}
+
 # Collect compose args (profiles and overrides) for an enabled provider.
 collect_provider_compose() {
     local i="$1"
@@ -645,6 +757,7 @@ DCM_KUBECONFIG="${KUBECONFIG:-}"
 OPENSHIFT_API="${OPENSHIFT_API:-}"
 OPENSHIFT_USERNAME="${OPENSHIFT_USERNAME:-kubeadmin}"
 OPENSHIFT_PASSWORD="${OPENSHIFT_PASSWORD:-}"
+AUTH_ENABLED_EXPLICIT=false
 COMPOSE_EXTRA_FILE_ARGS=()
 
 require_arg() {
@@ -727,6 +840,8 @@ while [[ $# -gt 0 ]]; do
             require_arg "$1" "${2:-}"
             COMPOSE_EXTRA_FILE_ARGS+=("-f" "$(cd "$(dirname "${2:-}")" && pwd)/$(basename "${2:-}")")
             shift 2 ;;
+        --auth-enabled)
+            AUTH_ENABLED_EXPLICIT=true; shift ;;
         --cleanup-on-failure)
             CLEANUP_ON_FAILURE=true; shift ;;
         --running-versions)
@@ -773,6 +888,23 @@ for i in $(seq 0 $((PROV_COUNT - 1))); do
     collect_provider_compose "${i}"
 done
 
+AUTH_ENABLED=false
+if [[ "${AUTH_ENABLED_EXPLICIT}" == true ]] || [[ "${AUTH_DISABLED:-}" == "false" ]]; then
+    AUTH_ENABLED=true
+fi
+
+# Read the existing deployment configuration for standalone inspection and teardown.
+# Do not source it: deploy/.env contains values that should not be executed as shell code.
+if [[ "${RUNNING_VERSIONS}" == true || "${TEAR_DOWN}" == true ]] &&
+    [[ -f "${CONTROL_PLANE_TMP_DIR}/deploy/.env" ]] &&
+    grep -Eq "^AUTH_DISABLED[[:space:]]*=[[:space:]]*(false|\"false\"|'false')[[:space:]]*$" "${CONTROL_PLANE_TMP_DIR}/deploy/.env"; then
+    AUTH_ENABLED=true
+fi
+
+if [[ "${AUTH_ENABLED}" == true ]]; then
+    COMPOSE_PROFILES+=("--profile" "auth")
+fi
+
 # --- Running versions (standalone) ----------------------------------------- #
 
 if [[ "${RUNNING_VERSIONS}" == true ]]; then
@@ -1076,6 +1208,8 @@ fi
 log "Cloning control-plane (repo=${CONTROL_PLANE_REPO}, branch=${CONTROL_PLANE_BRANCH})"
 git clone --branch "${CONTROL_PLANE_BRANCH}" --single-branch --depth 1 "${CONTROL_PLANE_REPO}" "${CONTROL_PLANE_TMP_DIR}"
 
+ensure_deploy_env "${CONTROL_PLANE_TMP_DIR}" || exit 1
+
 # --- Deploy ---------------------------------------------------------------- #
 
 if [[ "${CLEANUP_ON_FAILURE}" == true ]]; then
@@ -1090,6 +1224,9 @@ done
 if [[ ${#ENABLED_LABELS[@]} -gt 0 ]]; then
     info "Enabled providers: ${ENABLED_LABELS[*]}"
 fi
+if [[ "${AUTH_ENABLED}" == true ]]; then
+    info "Authentication enabled (compose profile: auth)"
+fi
 podman-compose -f "${CONTROL_PLANE_TMP_DIR}/deploy/compose.yaml" ${COMPOSE_EXTRA_FILE_ARGS[@]+"${COMPOSE_EXTRA_FILE_ARGS[@]}"} ${COMPOSE_PROFILES[@]+"${COMPOSE_PROFILES[@]}"} up -d
 
 echo

tests/run-e2e.sh

@@ -313,9 +313,9 @@ fi
 # Teardown the stack.
 if [[ "${SKIP_TEARDOWN}" == "false" ]]; then
     log "Tearing down DCM stack"
-    if ! "${DEPLOY_SCRIPT}" --tear-down; then
+    if ! "${DEPLOY_SCRIPT}" --tear-down "${DEPLOY_ARGS[@]+"${DEPLOY_ARGS[@]}"}"; then
         err "Teardown failed (non-fatal) — containers may still be running"
-        err "Manual cleanup: ${DEPLOY_SCRIPT} --tear-down"
+        err "Manual cleanup: ${DEPLOY_SCRIPT} --tear-down ${DEPLOY_ARGS[*]}"
     fi
 else
     log "Skipping teardown (--skip-teardown)"