fix(monitoring): fix production monitoring startup, alertmanager configuration, prometheus alert syntax, and mcp healthcheck

docker-compose.yml

@@ -150,7 +150,7 @@ services:
           cpus: '0.25'
           memory: 128M
     healthcheck:
-      test: ["CMD-SHELL", "wget --header \"x-api-key: ${MCP_API_KEY:-dev-key}\" -q -O /dev/null http://localhost:3001/ || exit 1"]
+      test: ["CMD-SHELL", "wget --header \"x-api-key: $${MCP_API_KEY:-dev-key}\" -q -O /dev/null http://localhost:3001/ || exit 1"]
       interval: 30s
       timeout: 10s
       retries: 3

monitoring/alerts.yml

@@ -37,7 +37,7 @@ groups:
           severity: warning
         annotations:
           summary: "⚠️ Disk space below 15%"
-          description: "Only {{ humanizePercentage (div (node_filesystem_avail_bytes{mountpoint='/'}) (node_filesystem_size_bytes{mountpoint='/'})) }} disk space remaining."
+          description: "Only {{ $value | humanizePercentage }} disk space remaining."
 
       - alert: DiskSpaceCritical
         expr: (node_filesystem_avail_bytes{mountpoint="/"} / node_filesystem_size_bytes{mountpoint="/"}) < 0.05

monitoring/docker-compose.monitoring.yml

@@ -55,11 +55,14 @@ services:
     image: metalmatze/alertmanager-bot:0.4.3
     container_name: memento-alertmanager-telegram
     restart: unless-stopped
+    profiles:
+      - telegram
     environment:
-      TELEGRAM_TOKEN: ${TELEGRAM_BOT_TOKEN:-}
+      TELEGRAM_TOKEN: ${TELEGRAM_BOT_TOKEN:-dummy_token}
-      TELEGRAM_ADMIN: ${TELEGRAM_CHAT_ID:-}
+      TELEGRAM_ADMIN: ${TELEGRAM_CHAT_ID:-0}
       ALERTMANAGER_URL: http://alertmanager:9093
-      STORE: /data/bolt.db
+      STORE: bolt
+      BOLT_PATH: /data/bolt.db
       LISTEN_ADDR: 0.0.0.0:8080
     volumes:
       - alertmanager-bot-data:/data
@@ -91,7 +94,7 @@ services:
     container_name: memento-postgres-exporter
     restart: unless-stopped
     env_file:
-      - /opt/memento/.env.docker
+      - ../.env.docker
     environment:
       DATA_SOURCE_NAME: "postgresql://${POSTGRES_USER:-memento}:${POSTGRES_PASSWORD:-memento}@memento-postgres:5432/${POSTGRES_DB:-memento}?sslmode=disable"
     ports:
@@ -141,4 +144,4 @@ networks:
     driver: bridge
   memento-net:
     external: true
-    name: memento_memento-network
+    name: ${MEMENTO_NETWORK_NAME:-memento_memento-network}

scripts/deploy-prod.sh

@@ -130,13 +130,7 @@ docker compose up -d --remove-orphans --force-recreate memento-note
 docker compose up -d memento-socket
 docker compose up -d mcp-server 2>/dev/null || true
 
-# Redémarrer les exporters monitoring pour appliquer les configs à jour
+# Monitoring stack updates are handled at the end of successful deployment
-if docker ps --format '{{.Names}}' | grep -q "^memento-grafana$"; then
-  echo "=== Updating monitoring exporters ==="
-  env $(cat /opt/memento/.env.docker | grep -v '^#' | xargs) \
-    docker compose -f monitoring/docker-compose.monitoring.yml up -d \
-      postgres-exporter cadvisor node-exporter redis-exporter 2>&1 || true
-fi
 
 echo "=== Migrations (Prisma CLI via node, pas npx) ==="
 if docker compose exec -T memento-note test -f ./node_modules/prisma/build/index.js 2>/dev/null; then
@@ -152,24 +146,25 @@ for i in $(seq 1 "$HEALTH_CHECK_MAX_ITERATIONS"); do
   ACTUAL=$(echo "$BODY" | jq -r '.commit // empty' 2>/dev/null || true)
   if [ "$ACTUAL" = "$GIT_COMMIT" ]; then
     echo "OK build-info commit=$ACTUAL"
-    if docker ps --format '{{.Names}}' | grep -q "^memento-grafana$"; then
+    echo "=== Updating monitoring stack ==="
-      # Ne recréer Prometheus que si sa config a changé (préserve l'historique TSDB)
+    if [ -f /opt/memento/.env.docker ]; then
-      PROM_CHANGED=$(git diff --name-only HEAD~1 HEAD 2>/dev/null | grep -E '^monitoring/(prometheus\.yml|alerts\.yml)' || true)
+      export $(cat /opt/memento/.env.docker | grep -v '^#' | xargs)
-      GRAFANA_CHANGED=$(git diff --name-only HEAD~1 HEAD 2>/dev/null | grep -E '^monitoring/' || true)
+    fi
-      if [ -n "$PROM_CHANGED" ]; then
+    if [ -n "${TELEGRAM_BOT_TOKEN:-}" ] && [ -n "${TELEGRAM_CHAT_ID:-}" ]; then
-        echo "=== Prometheus config changed — recreating ==="
+      echo "=== Starting Monitoring Stack (with Telegram bot) ==="
-        docker compose -f monitoring/docker-compose.monitoring.yml up -d --force-recreate prometheus
+      docker compose -f monitoring/docker-compose.monitoring.yml --profile telegram up -d --remove-orphans 2>&1 || echo "WARN: Failed to bring up monitoring stack"
-      else
+    else
-        echo "=== Prometheus config unchanged — keeping TSDB history ==="
+      echo "=== Starting Monitoring Stack (without Telegram bot) ==="
-        docker compose -f monitoring/docker-compose.monitoring.yml up -d prometheus
+      docker compose -f monitoring/docker-compose.monitoring.yml up -d --remove-orphans 2>&1 || echo "WARN: Failed to bring up monitoring stack"
-      fi
+    fi
-      if [ -n "$GRAFANA_CHANGED" ]; then
+
-        echo "=== Grafana config changed — recreating ==="
+    if docker ps --format '{{.Names}}' | grep -q "^memento-prometheus$"; then
-        docker compose -f monitoring/docker-compose.monitoring.yml up -d --force-recreate grafana
+      echo "=== Reloading Prometheus configuration ==="
-      else
+      docker compose -f monitoring/docker-compose.monitoring.yml exec -T prometheus kill -SIGHUP 1 2>/dev/null || true
-        echo "=== Grafana config unchanged — keeping state ==="
+    fi
-        docker compose -f monitoring/docker-compose.monitoring.yml up -d grafana
+    if docker ps --format '{{.Names}}' | grep -q "^memento-alertmanager$"; then
-      fi
+      echo "=== Reloading Alertmanager configuration ==="
+      docker compose -f monitoring/docker-compose.monitoring.yml exec -T alertmanager kill -SIGHUP 1 2>/dev/null || true
     fi
     docker compose ps
     telegram_notify "success" "Deployment successful — app is healthy"