Deploy automation: 8-step pipeline with volume mounting, multi-container docker-compose

This commit is contained in:
Agent Zero
2026-07-25 22:48:38 +02:00
parent 20e6545aa1
commit 745b634e7c
2 changed files with 132 additions and 70 deletions
+53 -56
View File
@@ -1,27 +1,30 @@
# =============================================================================
# docker-compose.yml — CRM System v1.0 (local testing with PostgreSQL)
# docker-compose.yml — LeoCRM Multi-Container Setup
#
# Use this file to run a full Postgres + crm-app stack locally, e.g. for
# smoke-testing the Docker image or running the backend against a real
# PostgreSQL instance before deploying to Coolify.
# Full stack: PostgreSQL + Redis + App + Worker, all with persistent volumes.
#
# Production deploys in Coolify use COOLIFY_SETUP.md (single-container app +
# a Coolify-managed Postgres database), NOT this file.
# This file serves two purposes:
# 1. Local development / testing: `docker compose up --build`
# 2. Reference for the production multi-container architecture
#
# Usage:
# Production deploys via Coolify use scripts/deploy.py which automates:
# - Coolify API build & deploy
# - Persistent volume mounting (patched into Coolify's generated compose)
# - Worker container startup
# - RLS policy enforcement
# - Health & domain verification
#
# Usage (local testing):
# cp .env.docker.example .env.docker
# $EDITOR .env.docker # fill AUTH_SECRET, POSTGRES_PASSWORD, ...
# $EDITOR .env.docker # fill SECRET_KEY, POSTGRES_PASSWORD, ...
# docker compose --env-file .env.docker up --build
# curl http://localhost:8000/health
# curl http://localhost:8000/api/v1/health
# =============================================================================
services:
# -------------------------------------------------------------------------
# PostgreSQL 16 (Alpine) — local Postgres for development / smoke tests.
# Coolify will provision its own managed Postgres database in production.
# -------------------------------------------------------------------------
# ── PostgreSQL 16 with pgvector ─────────────────────────────────────
postgres:
image: postgres:16-alpine
image: pgvector/pgvector:pg16
container_name: crm-postgres
restart: unless-stopped
environment:
@@ -32,7 +35,7 @@ services:
volumes:
- pgdata:/var/lib/postgresql/data
ports:
- "5432:5432" # local-only convenience; remove for CI / prod-like runs
- "5432:5432" # local-only; remove for prod-like runs
healthcheck:
test: ["CMD-SHELL", "pg_isready -U ${POSTGRES_USER:-crm_user} -d ${POSTGRES_DB:-crm_db}"]
interval: 10s
@@ -42,10 +45,25 @@ services:
networks:
- crm-net
# -------------------------------------------------------------------------
# CRM App — built from the local Dockerfile.
# NOTE: no `image:` directive — we build from source on `docker compose up`.
# -------------------------------------------------------------------------
# ── Redis 7 (sessions, rate limiting, ARQ queue) ────────────────────
redis:
image: redis:7-alpine
container_name: crm-redis
restart: unless-stopped
command: redis-server --requirepass ${REDIS_PASSWORD:-changeme}
volumes:
- redisdata:/data
ports:
- "6379:6379" # local-only; remove for prod-like runs
healthcheck:
test: ["CMD", "redis-cli", "-a", "${REDIS_PASSWORD:-changeme}", "ping"]
interval: 10s
timeout: 5s
retries: 5
networks:
- crm-net
# ── CRM API (FastAPI + Uvicorn) ─────────────────────────────────────
crm-app:
build:
context: .
@@ -58,17 +76,23 @@ services:
redis:
condition: service_healthy
environment:
# Use the internal docker-compose DNS name "postgres" (NOT localhost)
DATABASE_URL: ${DATABASE_URL:?DATABASE_URL is required}
REDIS_URL: ${REDIS_URL:-redis://:${REDIS_PASSWORD:-changeme}@redis:6379/0}
SECRET_KEY: ${SECRET_KEY:?SECRET_KEY is required (min 32 chars)}
# Frontend served from same origin in production; allow local dev hosts too
CORS_ORIGINS: ${CORS_ORIGINS:-http://localhost:8000,http://localhost:5173}
ENVIRONMENT: ${ENVIRONMENT:-production}
LOG_LEVEL: ${LOG_LEVEL:-INFO}
SESSION_COOKIE_SECURE: ${SESSION_COOKIE_SECURE:-true}
STORAGE_PATH: ${STORAGE_PATH:-/data/storage}
BCRYPT_ROUNDS: ${BCRYPT_ROUNDS:-12}
# S3 Storage (optional — if STORAGE_BACKEND=s3)
STORAGE_BACKEND: ${STORAGE_BACKEND:-local}
S3_ENDPOINT: ${S3_ENDPOINT:-}
S3_BUCKET: ${S3_BUCKET:-}
S3_ACCESS_KEY: ${S3_ACCESS_KEY:-}
S3_SECRET_KEY: ${S3_SECRET_KEY:-}
S3_REGION: ${S3_REGION:-us-east-1}
S3_SECURE: ${S3_SECURE:-true}
ports:
- "8000:8000"
volumes:
@@ -82,61 +106,34 @@ services:
networks:
- crm-net
# -------------------------------------------------------------------------
# CRM Worker — ARQ background worker (same image, different entrypoint).
# Runs migrations? No — the API container handles migrations.
# Scale with `docker compose up --scale crm-worker=N`.
# Cron jobs use a Redis-based distributed lock so only one replica fires.
# -------------------------------------------------------------------------
# ── CRM Worker (ARQ background jobs, cron, outbox processor) ───────
crm-worker:
build:
context: .
dockerfile: Dockerfile
container_name: crm-worker
restart: unless-stopped
entrypoint: ["/app/worker.sh"]
depends_on:
postgres:
condition: service_healthy
redis:
condition: service_healthy
entrypoint: ["/app/worker.sh"]
environment:
DATABASE_URL: ${DATABASE_URL:?DATABASE_URL is required}
REDIS_URL: ${REDIS_URL:-redis://:${REDIS_PASSWORD:-changeme}@redis:6379/0}
SECRET_KEY: ${SECRET_KEY:?SECRET_KEY is required (min 32 chars)}
ENVIRONMENT: ${ENVIRONMENT:-production}
LOG_LEVEL: ${LOG_LEVEL:-INFO}
SESSION_COOKIE_SECURE: ${SESSION_COOKIE_SECURE:-true}
STORAGE_PATH: ${STORAGE_PATH:-/data/storage}
BCRYPT_ROUNDS: ${BCRYPT_ROUNDS:-12}
STORAGE_BACKEND: ${STORAGE_BACKEND:-local}
S3_ENDPOINT: ${S3_ENDPOINT:-}
S3_BUCKET: ${S3_BUCKET:-}
S3_ACCESS_KEY: ${S3_ACCESS_KEY:-}
S3_SECRET_KEY: ${S3_SECRET_KEY:-}
volumes:
- storage:/data/storage
healthcheck:
# Check if the ARQ worker process is alive
test: ["CMD-SHELL", "pgrep -f \"arq app.core.worker.WorkerSettings\" || exit 1"]
interval: 30s
timeout: 10s
retries: 3
start_period: 15s
networks:
- crm-net
# -------------------------------------------------------------------------
# Redis 7 (Alpine) — sessions, rate limiting, ARQ queue.
# -------------------------------------------------------------------------
redis:
image: redis:7-alpine
container_name: crm-redis
restart: unless-stopped
command: redis-server --requirepass ${REDIS_PASSWORD:-changeme}
volumes:
- redisdata:/data
ports:
- "6379:6379" # local-only convenience; remove for prod-like runs
healthcheck:
test: ["CMD", "redis-cli", "-a", "${REDIS_PASSWORD:-changeme}", "ping"]
interval: 10s
timeout: 5s
retries: 5
networks:
- crm-net
Executable → Regular
+79 -14
View File
@@ -280,6 +280,61 @@ def verify_db() -> DeployResult:
return DeployResult(all_pass, "DB verification " + ("passed" if all_pass else "failed"), 0, details)
def ensure_volume() -> DeployResult:
"""Ensure persistent volume is mounted on the app container.
Coolify regenerates docker-compose.yaml on each deploy, overwriting manual
volume config. This function patches the generated compose file to add the
volume, then restarts the container.
"""
print(" Ensuring persistent volume...")
# Ensure volume exists
code, output = ssh_run("docker volume inspect leocrm-storage >/dev/null 2>&1 || docker volume create leocrm-storage")
if code != 0:
return DeployResult(False, f"Failed to ensure volume: {output}")
# Patch the Coolify-generated docker-compose.yaml to add volume
patch_script = '''python3 -c "
import sys
path = '/data/coolify/applications/stvabl4vaqru7jclx4ittzr3/docker-compose.yaml'
with open(path) as f:
content = f.read()
# Add volume to service section (after env_file)
old = ' env_file:\\n - .env\\n'
new = ' env_file:\\n - .env\\n volumes:\\n - leocrm-storage:/data/storage\\n'
if 'leocrm-storage:/data/storage' not in content:
content = content.replace(old, new, 1)
# Add top-level volumes section if missing
if 'volumes:' not in content.split('networks:')[0] if 'networks:' in content else True:
if 'volumes:\\n leocrm-storage:' not in content:
content += '\\nvolumes:\\n leocrm-storage:\\n external: true\\n'
with open(path, 'w') as f:
f.write(content)
print('Volume patched')
"'''
code, output = ssh_run(patch_script)
if code != 0:
return DeployResult(False, f"Failed to patch compose file: {output}")
# Restart container with volume
code, output = ssh_run("cd /data/coolify/applications/stvabl4vaqru7jclx4ittzr3/ && docker compose down 2>&1 && docker compose up -d 2>&1")
if code != 0:
return DeployResult(False, f"Failed to restart with volume: {output}")
# Verify volume is mounted
import time as _time
_time.sleep(5)
code, output = ssh_run('docker inspect $(docker ps --format "{{.Names}}" | grep stvabl4 | head -1) --format "{{json .Mounts}}" 2>/dev/null')
if code == 0 and "leocrm-storage" in output:
return DeployResult(True, "Persistent volume mounted")
return DeployResult(False, f"Volume not found in mounts: {output}")
def ensure_rls() -> DeployResult:
"""Ensure RLS is active on all tenant tables (idempotent)."""
print(" Ensuring RLS on all tenant tables...")
@@ -336,7 +391,7 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
# Step 1: Trigger Coolify deploy
if not skip_build and not migrate_only:
print("\n[1/7] Triggering Coolify build & deploy...")
print("\n[1/8] Triggering Coolify build & deploy...")
try:
result = client.deploy(APP_UUID)
deploy_info = result["deployments"][0]
@@ -357,9 +412,19 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
print("\n[1/7] Skipping build (skip-build flag)")
steps.append(("Coolify deploy", DeployResult(True, "Skipped")))
# Step 2: Wait for healthy container
# Step 2: Ensure persistent volume
if not migrate_only:
print("\n[2/7] Waiting for container health...")
print("\n[2/8] Ensuring persistent volume...")
vol_result = ensure_volume()
steps.append(("Persistent volume", vol_result))
if not vol_result.success:
print(f" ⚠️ {vol_result.message} (non-fatal)")
else:
print(f"{vol_result.message}")
# Step 3: Wait for healthy container
if not migrate_only:
print("\n[3/8] Waiting for container health...")
health_result = wait_for_healthy(timeout=120)
steps.append(("Container health", health_result))
if not health_result.success:
@@ -367,8 +432,8 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
return 1
print(f"{health_result.message}")
# Step 3: Ensure RLS
print("\n[3/7] Ensuring RLS policies...")
# Step 4: Ensure RLS
print("\n[4/8] Ensuring RLS policies...")
rls_result = ensure_rls()
steps.append(("RLS policies", rls_result))
if not rls_result.success:
@@ -376,8 +441,8 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
else:
print(f"{rls_result.message}")
# Step 4: Verify DB
print("\n[4/7] Verifying database...")
# Step 5: Verify DB
print("\n[5/8] Verifying database...")
db_result = verify_db()
steps.append(("DB verification", db_result))
if db_result.success:
@@ -387,9 +452,9 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
else:
print(f" ⚠️ {db_result.message}")
# Step 5: Start worker
# Step 6: Start worker
if not migrate_only:
print("\n[5/7] Starting worker container...")
print("\n[6/8] Starting worker container...")
worker_result = start_worker()
steps.append(("Worker", worker_result))
if not worker_result.success:
@@ -397,8 +462,8 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
else:
print(f"{worker_result.message}")
# Step 6: Verify health
print("\n[6/7] Verifying app health...")
# Step 7: Verify health
print("\n[7/8] Verifying app health...")
app_health = verify_health()
steps.append(("App health", app_health))
if not app_health.success:
@@ -406,8 +471,8 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
return 1
print(f"{app_health.message}")
# Step 7: Verify domain
print("\n[7/7] Verifying domain...")
# Step 8: Verify domain
print("\n[8/8] Verifying domain...")
domain_result = verify_domain(domain)
steps.append(("Domain", domain_result))
if domain_result.success:
@@ -424,7 +489,7 @@ def deploy(environment: str = "production", skip_build: bool = False, migrate_on
print(f" {status} {name}: {result.message}")
# Return success if all critical steps passed
critical = [s for s in steps if s[0] in ("Coolify deploy", "Container health", "App health")]
critical = [s for s in steps if s[0] in ("Coolify deploy", "Container health", "App health", "Persistent volume")]
all_critical = all(s[1].success for s in critical)
print(f"\n Overall: {'✅ SUCCESS' if all_critical else '❌ FAILED'}\n")
return 0 if all_critical else 1