Fix deployment script permission denied on marker files (#5272)

The saves directory is owned by root (created by Docker) but the deploy
script runs as the deploy user. Use docker exec to create marker files
from inside a running container that has the saves directory mounted.

- create_deployment_marker: uses active container (running before staging starts)
- create_flush_marker: uses staging container (running after active stops)
- cleanup_markers_on_failure: uses active container (still running on failure)
- remove_stale_deployment_marker: finds any running eagle container

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

scripts/deploy-blue-green.sh

@@ -46,6 +46,37 @@ log_info() { echo -e "${GREEN}[INFO]${NC} $1"; }
 log_warn() { echo -e "${YELLOW}[WARN]${NC} $1"; }
 log_error() { echo -e "${RED}[ERROR]${NC} $1"; }
 
+# Marker file operations use docker exec because saves directory is owned by root (Docker).
+# We run commands inside a container that has the saves directory mounted.
+create_deployment_marker() {
+    local deploy_id=$1
+    local container=$2  # Container to use for file operations
+    docker exec "${container}" rm -f /app/saves/.flush_complete
+    docker exec "${container}" sh -c "echo '${deploy_id}' > /app/saves/.deployment_in_progress"
+}
+
+create_flush_marker() {
+    local deploy_id=$1
+    local container=$2  # Container to use for file operations
+    docker exec "${container}" sh -c "echo '${deploy_id}' > /app/saves/.flush_complete"
+    docker exec "${container}" rm -f /app/saves/.deployment_in_progress
+}
+
+cleanup_markers_on_failure() {
+    local container=$1  # Container to use for file operations
+    docker exec "${container}" rm -f /app/saves/.deployment_in_progress 2>/dev/null || true
+    docker exec "${container}" touch /app/saves/.flush_complete 2>/dev/null || true
+}
+
+remove_stale_deployment_marker() {
+    # Try any running eagle container
+    local container
+    container=$(docker ps --filter "name=eagle-" --format "{{.Names}}" | head -1)
+    if [ -n "${container}" ]; then
+        docker exec "${container}" rm -f /app/saves/.deployment_in_progress 2>/dev/null || true
+    fi
+}
+
 # Determine which instance is currently running (not from nginx config)
 get_running_instance() {
     local blue_running green_running
@@ -154,13 +185,7 @@ main() {
 
     cd "${APP_DIR}"
 
-    # Step 1: Signal deployment in progress
-    log_info "[DEPLOY:${deploy_id}] Step 1: Signaling deployment in progress..."
-    rm -f "${FLUSH_MARKER}"
-    echo "${deploy_id}" > "${DEPLOYMENT_IN_PROGRESS}"
-    log_info "[DEPLOY:${deploy_id}] Deployment marker created"
-
-    # Determine current active instance
+    # Determine current active instance (need this before creating marker)
     local active=$(get_running_instance)
     local staging
     if [ "$active" = "blue" ] || [ "$active" = "none" ]; then
@@ -170,15 +195,20 @@ main() {
         staging="blue"
     fi
 
-    log_info "Active instance: eagle-${active}"
-    log_info "Staging instance: eagle-${staging}"
+    # Step 1: Signal deployment in progress
+    log_info "[DEPLOY:${deploy_id}] Step 1: Signaling deployment in progress..."
+    # Use active container for marker operations (it's the one currently running)
+    if [ "$active" != "none" ] && docker ps --filter "name=eagle-${active}" --format "{{.Names}}" | grep -q .; then
+        create_deployment_marker "${deploy_id}" "eagle-${active}"
+        log_info "[DEPLOY:${deploy_id}] Deployment marker created via eagle-${active}"
+    else
+        log_warn "[DEPLOY:${deploy_id}] No running container to create marker (first deploy?)"
+    fi
 
     # Pull the new image (with retry for intermittent registry issues)
     if ! pull_with_retry "${new_image}" 3; then
         log_error "Failed to pull new image, aborting deployment"
-        # Clean up deployment markers on failure
-        rm -f "${DEPLOYMENT_IN_PROGRESS}"
-        touch "${FLUSH_MARKER}"
+        cleanup_markers_on_failure "eagle-${active}"
         exit 1
     fi
 
@@ -194,9 +224,7 @@ main() {
     if ! wait_for_healthy "eagle-${staging}" 90; then
         log_error "Staging instance failed health check, aborting deployment"
         docker compose -f "${COMPOSE_FILE}" stop "eagle-${staging}"
-        # Clean up deployment markers on failure
-        rm -f "${DEPLOYMENT_IN_PROGRESS}"
-        touch "${FLUSH_MARKER}"
+        cleanup_markers_on_failure "eagle-${active}"
         exit 1
     fi
 
@@ -213,9 +241,7 @@ main() {
         if ! "${WARMUP_SCRIPT}" "localhost:${staging_port}"; then
             log_error "Warmup/smoke test failed, aborting deployment"
             docker compose -f "${COMPOSE_FILE}" stop "eagle-${staging}"
-            # Clean up deployment markers on failure
-            rm -f "${DEPLOYMENT_IN_PROGRESS}"
-            touch "${FLUSH_MARKER}"
+            cleanup_markers_on_failure "eagle-${active}"
             exit 1
         fi
     else
@@ -279,8 +305,7 @@ main() {
     # Step 6: Create flush marker - signals that disk state is fresh
     # Any waiting lazy-loads on staging will now proceed with fresh data.
     log_info "[DEPLOY:${deploy_id}] Step 6: Creating flush marker..."
-    echo "${deploy_id}" > "${FLUSH_MARKER}"
-    rm -f "${DEPLOYMENT_IN_PROGRESS}"
+    create_flush_marker "${deploy_id}" "eagle-${staging}"
     log_info "[DEPLOY:${deploy_id}] Flush marker created - waiting lazy-loads can now proceed"
 
     # Update .env for admin service
@@ -416,7 +441,7 @@ check_requirements() {
     # Clean up any stale deployment-in-progress marker from a previous failed deploy
     if [ -f "${DEPLOYMENT_IN_PROGRESS}" ]; then
         log_warn "Found stale deployment-in-progress marker, removing it"
-        rm -f "${DEPLOYMENT_IN_PROGRESS}"
+        remove_stale_deployment_marker
     fi
 }