stability and vps runtime updates

docker-compose.yml

@@ -1,5 +1,5 @@
 x-app-env: &app-env
-  # Всегда используем Postgres.
+  # NB: hardcoded to Postgres — .env may contain sqlite for local CLI, don't let it leak here
   IAAI_DATABASE_URL: postgresql+psycopg2://iaai:iaai@postgres:5432/iaai_scraper
   IAAI_REDIS_URL: ${IAAI_REDIS_URL:-redis://redis:6379/0}
   CELERY_BROKER_URL: ${CELERY_BROKER_URL:-redis://redis:6379/0}
@@ -8,7 +8,8 @@ x-app-env: &app-env
   CELERY_TASK_SOFT_TIME_LIMIT: ${CELERY_TASK_SOFT_TIME_LIMIT:-900}
   CELERY_TASK_TIME_LIMIT: ${CELERY_TASK_TIME_LIMIT:-1200}
   CELERY_BROKER_VISIBILITY_TIMEOUT: ${CELERY_BROKER_VISIBILITY_TIMEOUT:-2400}
-  # 0 = без лимита.
+  # 0 => без лимита (в коде интерпретируется как None).
+  # После первичного полного прохода hourly-режим должен успевать за всеми новыми авто.
   CELERY_BEAT_SYNC_LIMIT: ${CELERY_BEAT_SYNC_LIMIT:-0}
   CELERY_WORKER_MAX_TASKS_PER_CHILD: ${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
   CELERY_BATCH_SIZE: ${CELERY_BATCH_SIZE:-200}
@@ -24,7 +25,8 @@ x-app-env: &app-env
   IAAI_TOKENS_FILE: ${IAAI_TOKENS_FILE:-/data/tokens.json}
   IAAI_RUNTIME_CONFIG_FILE: ${IAAI_RUNTIME_CONFIG_FILE:-/app/runtime_config.json}
   IAAI_BROWSER_ENGINE: chromium
-  # Self-heal для worker.
+  # Self-heal watchdog (внутри worker): если очередь есть, а глобального прогресса долго нет,
+  # watchdog перезапускает только зависший worker-процесс.
   IAAI_SELF_HEAL_ENABLED: ${IAAI_SELF_HEAL_ENABLED:-true}
   IAAI_SELF_HEAL_CHECK_INTERVAL_SECONDS: ${IAAI_SELF_HEAL_CHECK_INTERVAL_SECONDS:-30}
   IAAI_SELF_HEAL_STALL_SECONDS: ${IAAI_SELF_HEAL_STALL_SECONDS:-900}
@@ -50,7 +52,7 @@ x-worker-service: &worker-service
     - tokens_data:/data
 
 services:
-  # PostgreSQL.
+  # PostgreSQL
   postgres:
     image: postgres:16-alpine
     container_name: iaai-postgres
@@ -74,7 +76,7 @@ services:
         max-size: "10m"
         max-file: "5"
 
-  # Redis.
+  # Redis (Celery broker)
   redis:
     image: redis:7-alpine
     container_name: iaai-redis
@@ -98,7 +100,7 @@ services:
         max-size: "10m"
         max-file: "5"
 
-  # Миграции.
+  # DB migrations
   migrate:
     <<: *app-service
     container_name: iaai-migrate
@@ -108,7 +110,7 @@ services:
         condition: service_healthy
     command: alembic upgrade head
 
-  # API.
+  # FastAPI
   api:
     <<: *app-service
     container_name: iaai-api
@@ -136,7 +138,7 @@ services:
         max-size: "10m"
         max-file: "5"
 
-  # Worker.
+  # Celery Worker
   worker:
     <<: *worker-service
     restart: unless-stopped
@@ -153,7 +155,7 @@ services:
       --pidfile=/tmp/celery-worker.pid
       -Q scraping --max-tasks-per-child=${CELERY_WORKER_MAX_TASKS_PER_CHILD:-5}
     healthcheck:
-      # Проверка worker.
+      # Worker считается healthy только если живы и Celery PID, и self-heal watchdog.
       test: ["CMD-SHELL", "test -f /tmp/celery-worker.pid && kill -0 $(cat /tmp/celery-worker.pid) && pgrep -f 'iaai_scraper.worker.self_heal' >/dev/null"]
       interval: 60s
       timeout: 10s
@@ -165,7 +167,7 @@ services:
         max-size: "10m"
         max-file: "5"
 
-  # Beat.
+  # Celery Beat (периодический планировщик)
   beat:
     <<: *worker-service
     container_name: iaai-beat