Files
ERP/docker-compose.yml
T
whqreza cb621ca5d0 feat(monitoring): multi-model dashboard v2.1 — tabs, per-model records, voice playback + MinIO dual-write
Problem: the dashboard was Jev-centric; NLU/arbiter/fallback DeepSeek, STT
providers and the audio-enhancer were barely visible, and stored texts/voices
had no real access path.

Bot (additive; reply path untouched):
- nlu.result now records latency_ms + the DeepSeek usage tokens
- matching.arbitration/fallback.result fill the model + latency_ms columns
  (payload unchanged; SQL coalesces column/payload so old rows still count)
- storage.py: MinIO dual-write layer — every voice (raw + denoised) and user
  attachment is mirrored fire-and-forget into bucket bot-media; the blocking
  minio SDK runs via asyncio.to_thread (senior-engineer review) so the event
  loop is never stalled; empty MINIO_* env = layer off; bucket auto-created
- requirements: minio>=7.2

Backend (admin-gated, read-only):
- GET /bot-metrics/models — one record per model (jev, deepseek-nlu,
  deepseek-arbiter, deepseek-fallback, audio-enhancer + dynamic per-provider
  STT) with calls/ok-rate/p50/p95/tokens/last-seen + version-drift vs the pin
- GET /bot-metrics/models/{id}/daily — per-model daily calls/fails/tokens/p50
- GET /bot-metrics/stages — every stage tagged with its Persian pipeline
  section (صوت/درک پیام/تطبیق/ابهام و انتخاب/پیش‌نویس)
- /traces now carries input_kind + a 120-char text excerpt (consensus →
  input.text → first STT transcript)
- media serving: POST /media-url mints a short-lived HMAC URL (audio/a tags
  cannot send Authorization); GET /media streams from MinIO bot-media with
  the read-only ./bot/data mount as fallback for pre-MinIO files; hard ref
  validation (voices|attachments only, no .., extension whitelist)
- router exempted from the global Bearer dependency (files.py pattern) —
  every endpoint keeps its own require_role('admin')

Frontend /bot-monitor (RTL, recharts):
- four tabs: نمای کلی / مدل‌ها (records table + per-model expandable daily
  charts: calls/fails, tokens, p50) / بخش‌ها (sections grouped) / پیام‌ها
- message drawer rebuilt: inline <audio controls> for raw + denoised voice,
  download links, every STT transcript + consensus side-by-side, user
  attachments, then the event timeline
- terminology: «ترس» → «پرونده‌ی پیام» with tooltips + a one-line legend

Grafana: bot-observability gains a «کارکرد مدل‌ها» table (stage→model
mapping for legacy rows); the explorer gains a message text column.

Seed: real (tiny) ogg files written + mirrored to MinIO, NLU usage tokens,
column-level model/latency on matching events.

Verified locally: /models returns all 7 records incl. per-STT and NLU tokens;
media 200 via signed URL (MinIO) and 401 without; ../ and .py rejected (400);
disk fallback serves a legacy file; pm-role 403 on every new endpoint;
bucket auto-created with 14 mirrored objects; tsc + vite build clean;
Grafana re-provisions 24 panels.
2026-10-25 22:02:40 +03:30

381 lines
14 KiB
YAML

services:
postgres:
image: postgres:18
container_name: procurement-postgres
environment:
POSTGRES_DB: procurement
POSTGRES_USER: postgres
# Local dev default; production overrides via .env (required by the
# prod overlay) — never commit real passwords.
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:-postgres}
ports:
- "5455:5432"
networks:
- default
- shared # monitoring stack scrapes it (postgres-exporter)
volumes:
- postgres_data:/var/lib/postgresql
- ./sql/migration.sql:/docker-entrypoint-initdb.d/01-migration.sql:ro
- ./sql/02-casdoor-db.sql:/docker-entrypoint-initdb.d/02-casdoor-db.sql:ro
healthcheck:
test: ["CMD-SHELL", "pg_isready -U postgres -d procurement"]
interval: 5s
timeout: 5s
retries: 10
restart: unless-stopped
mem_limit: 1g
# ==========================================================================
# Traefik v3 — public reverse proxy (replaces nginx as the edge)
# App is served on http://localhost (port 80) instead of :3000.
# Production: see docker-compose.prod.yml (real domains + HTTPS via
# Let's Encrypt, HTTP→HTTPS redirect).
# ==========================================================================
traefik:
# v3.6+ required: Traefik ≤3.3 fails against Docker Engine 29.x
# ("Error response from daemon" — Docker API version mismatch).
image: traefik:v3.6
container_name: procurement-traefik
restart: unless-stopped
mem_limit: 128m
command:
- --providers.docker=true
- --providers.docker.exposedbydefault=false
# Containers are on several networks; Traefik must use the dedicated
# proxy network IP (otherwise it may pick the default-network IP → 504).
- --providers.docker.network=traefik_network
- --providers.file.filename=/etc/traefik/dynamic.yml
- --entrypoints.web.address=:80
- --entrypoints.websecure.address=:443
- --certificatesresolvers.letsencrypt.acme.email=${TRAEFIK_ACME_EMAIL:-admin@example.com}
- --certificatesresolvers.letsencrypt.acme.storage=/certs/acme.json
- --certificatesresolvers.letsencrypt.acme.httpchallenge=true
- --certificatesresolvers.letsencrypt.acme.httpchallenge.entrypoint=web
- --log.level=INFO
ports:
- "80:80"
- "443:443"
volumes:
# NOTE: mounted read-write — Docker Desktop rejects read-only
# /var/run/docker.sock mounts ("Error response from daemon").
- /var/run/docker.sock:/var/run/docker.sock
- ./infra/traefik-dynamic.yml:/etc/traefik/dynamic.yml:ro
- traefik_certs:/certs
networks:
- traefik_network
minio:
image: minio/minio:latest
container_name: procurement-minio
command: server /data --console-address ":9001"
environment:
MINIO_ROOT_USER: ${MINIO_ROOT_USER:-minioadmin}
MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:-minioadmin}
ports:
- "9000:9000" # S3 API
- "9001:9001" # Web console
volumes:
- minio_data:/data
labels:
- traefik.enable=true
# S3 API
- traefik.http.routers.minio-api.rule=Host(`${S3_HOST:-s3.localhost}`)
- traefik.http.routers.minio-api.entrypoints=web
- traefik.http.routers.minio-api.service=minio-api-svc
- traefik.http.services.minio-api-svc.loadbalancer.server.port=9000
# Web console
- traefik.http.routers.minio-console.rule=Host(`${STORAGE_HOST:-storage.localhost}`)
- traefik.http.routers.minio-console.entrypoints=web
- traefik.http.routers.minio-console.service=minio-console-svc
- traefik.http.services.minio-console-svc.loadbalancer.server.port=9001
networks:
- traefik_network
- default
restart: unless-stopped
mem_limit: 512m
# MinIO Client (mc): waits for the server, creates the bucket, exits.
# Use for ad-hoc ops too: docker compose run --rm minio-init ...
minio-init:
image: minio/mc:latest
container_name: procurement-minio-init
depends_on:
minio:
condition: service_started
environment:
MINIO_ENDPOINT: minio:9000
MINIO_ROOT_USER: ${MINIO_ROOT_USER:-minioadmin}
MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:-minioadmin}
MINIO_BUCKET: procurement-files
entrypoint: >
/bin/sh -c "
until mc alias set local http://$${MINIO_ENDPOINT} $${MINIO_ROOT_USER} $${MINIO_ROOT_PASSWORD}; do
echo 'waiting for minio server...'; sleep 2;
done;
mc mb --ignore-existing local/$${MINIO_BUCKET};
echo bucket $${MINIO_BUCKET} ready;
"
restart: "no"
# ==========================================================================
# Scheduled logical backups of BOTH databases (procurement + casdoor).
# Writes pg_dump -Fc archives to a bind mount (default ./backups, point
# BACKUP_DIR at /srv/erp/backups on the VPS) and prunes archives older
# than BACKUP_KEEP_DAYS. Runs on the same postgres:18 image already pulled.
# ==========================================================================
db-backup:
image: postgres:18
container_name: procurement-db-backup
depends_on:
postgres:
condition: service_healthy
environment:
PGPASSWORD: ${POSTGRES_PASSWORD:-postgres}
BACKUP_INTERVAL_SECONDS: ${BACKUP_INTERVAL_SECONDS:-86400}
BACKUP_KEEP_DAYS: ${BACKUP_KEEP_DAYS:-14}
volumes:
- ${BACKUP_DIR:-./backups}:/backups
entrypoint: >
/bin/sh -c "
echo 'db-backup: dumping every $${BACKUP_INTERVAL_SECONDS}s, keeping $${BACKUP_KEEP_DAYS} days';
while true; do
STAMP=$$(date +%Y%m%d-%H%M%S);
pg_dump -h postgres -U postgres -d procurement -Fc -f /backups/procurement-$${STAMP}.dump &&
echo \"db-backup: procurement -> /backups/procurement-$${STAMP}.dump\";
pg_dump -h postgres -U postgres -d casdoor -Fc -f /backups/casdoor-$${STAMP}.dump &&
echo \"db-backup: casdoor -> /backups/casdoor-$${STAMP}.dump\";
find /backups -name '*.dump' -mtime +$${BACKUP_KEEP_DAYS} -delete;
sleep $${BACKUP_INTERVAL_SECONDS};
done
"
restart: unless-stopped
mem_limit: 256m
# ==========================================================================
# v2.6 — Casdoor (OIDC identity provider)
# Console: http://localhost:8080 (admin / 123)
# Application "inventory-app" (org "inventory", client inventory-app-client)
# is created by backend/tools/seed_casdoor.py — run it once after the
# first boot:
# docker compose run --rm -e CASDOOR_BASE=http://casdoor:8000 api python tools/seed_casdoor.py
#
# Casdoor uses the SAME postgres instance as the app (database `casdoor`,
# created by sql/02-casdoor-db.sql on a fresh volume).
# ==========================================================================
casdoor:
image: casbin/casdoor:latest
container_name: procurement-casdoor
depends_on:
postgres:
condition: service_healthy
environment:
RUNNING_IN_DOCKER: "true"
driverName: postgres
dataSourceName: "user=casdoor password=casdoor host=postgres port=5432 sslmode=disable dbname=casdoor"
dbName: casdoor
origin: "http://localhost:8080"
ports:
- "8080:8000"
labels:
- traefik.enable=true
- traefik.http.routers.casdoor.rule=Host(`${AUTH_HOST:-auth.localhost}`)
- traefik.http.routers.casdoor.entrypoints=web
- traefik.http.services.casdoor.loadbalancer.server.port=8000
networks:
- traefik_network
- default
restart: unless-stopped
mem_limit: 768m
api:
image: v15-api:latest
build:
context: ./backend
dockerfile: Dockerfile
container_name: procurement-api
volumes:
# Bot media archive (voices/attachments) — READ-ONLY fallback for the
# admin media endpoint when an object is not (yet) in MinIO.
- ./bot/data:/srv/bot/data:ro
environment:
PROCUREMENT_DATABASE_URL: postgresql+asyncpg://postgres:${POSTGRES_PASSWORD:-postgres}@postgres:5432/procurement
PROCUREMENT_MINIO_ENDPOINT: minio:9000
PROCUREMENT_MINIO_ACCESS_KEY: ${MINIO_ROOT_USER:-minioadmin}
PROCUREMENT_MINIO_SECRET_KEY: ${MINIO_ROOT_PASSWORD:-minioadmin}
PROCUREMENT_MINIO_BUCKET: procurement-files
PROCUREMENT_BOT_MEDIA_BUCKET: ${BOT_MINIO_BUCKET:-bot-media}
PROCUREMENT_MINIO_SECURE: "false"
# "dev" enables the local defaults; "prod" turns on the fail-fast
# startup checks (no wildcard CORS, no default credentials).
PROCUREMENT_ENVIRONMENT: ${ENVIRONMENT:-dev}
PROCUREMENT_DEV_LOGIN_ENABLED: ${DEV_LOGIN_ENABLED:-false}
# Comma-separated browser origins (JSON-free .env friendly).
PROCUREMENT_CORS_ORIGINS: ${CORS_ORIGINS:-http://localhost,http://localhost:3000,http://localhost:5173}
# Casdoor: endpoint is the in-network service URL; the issuer must match
# the origin Casdoor advertises to browsers (its OIDC discovery doc).
PROCUREMENT_CASDOOR_ENDPOINT: http://casdoor:8000
PROCUREMENT_CASDOOR_ISSUER: ${CASDOOR_ISSUER:-http://localhost:8080}
PROCUREMENT_CASDOOR_CLIENT_ID: inventory-app-client
PROCUREMENT_CASDOOR_CLIENT_SECRET: ${CASDOOR_CLIENT_SECRET:-inventory-app-secret-2026}
PROCUREMENT_CASDOOR_ORGANIZATION: inventory
PROCUREMENT_CASDOOR_APP_NAME: inventory-app
PROCUREMENT_CASDOOR_AUDIENCE: inventory-app-client
# Console admin used by POST /users/sync (identity mirroring).
PROCUREMENT_CASDOOR_ADMIN_USER: admin
PROCUREMENT_CASDOOR_ADMIN_PASSWORD: ${CASDOOR_ADMIN_PASSWORD:-123}
# Observability (Phase 2) — see monitoring/README.md
PROCUREMENT_LOG_FORMAT: ${LOG_FORMAT:-text}
PROCUREMENT_SENTRY_DSN: ${SENTRY_DSN_BACKEND:-}
PROCUREMENT_OPENPANEL_API_URL: ${OPENPANEL_API_URL:-}
PROCUREMENT_OPENPANEL_CLIENT_ID: ${OPENPANEL_CLIENT_ID:-}
PROCUREMENT_OPENPANEL_CLIENT_SECRET: ${OPENPANEL_CLIENT_SECRET:-}
ports:
- "8000:8000"
networks:
- traefik_network
- default
- shared # monitoring stack scrapes api:8000/metrics
healthcheck:
test: ["CMD-SHELL", "python -c \"import urllib.request as u; u.urlopen('http://127.0.0.1:8000/health', timeout=3)\""]
interval: 30s
timeout: 5s
retries: 5
start_period: 40s
labels:
- traefik.enable=true
- traefik.http.routers.api.rule=Host(`${API_HOST:-api.localhost}`)
- traefik.http.routers.api.entrypoints=web
- traefik.http.services.api.loadbalancer.server.port=8000
depends_on:
postgres:
condition: service_healthy
minio-init:
condition: service_completed_successfully
restart: unless-stopped
mem_limit: 768m
frontend:
image: v15-frontend:latest
build:
context: ./frontend
args:
# Developer Mode panel on the login page — dev-only aid. Production
# builds MUST leave this false (identity comes from Casdoor tokens).
DEV_LOGIN_ENABLED: ${DEV_LOGIN_ENABLED:-false}
# Registry overridable for networks where npmmirror is unreachable.
NPM_REGISTRY: ${NPM_REGISTRY:-https://registry.npmmirror.com}
# Observability (build-time baked; empty = disabled)
VITE_OPENPANEL_URL: ${OPENPANEL_PUBLIC_URL:-}
VITE_OPENPANEL_CLIENT_ID: ${OPENPANEL_CLIENT_ID:-}
VITE_SENTRY_DSN: ${SENTRY_DSN_FRONTEND:-}
container_name: procurement-frontend
# از entrypoint پیش‌فرض image nginx رد می‌شویم: آن اسکریپت `apk manifest`
# اجرا می‌کند و اگر دسترسی به مخزن بسته‌ها نباشد بی‌نهایت معلق می‌ماند
# (سرویس بالا نمی‌آید و traefik خطای 502 می‌دهد). کانفیگ ما مستقل است و
# نیازی به آن اسکریپت‌ها ندارد.
entrypoint: ["nginx", "-g", "daemon off;"]
labels:
- traefik.enable=true
- traefik.http.routers.frontend.rule=Host(`${APP_HOST:-localhost}`)
- traefik.http.routers.frontend.entrypoints=web
- traefik.http.services.frontend.loadbalancer.server.port=80
networks:
- traefik_network
- default
depends_on:
- api
restart: unless-stopped
mem_limit: 128m
bot:
image: v15-bot:latest
build:
context: ./bot
dockerfile: Dockerfile
container_name: procurement-bot
volumes:
- ./bot/app:/srv/bot/app
- ./bot/tools:/srv/bot/tools
- ./bot/data:/srv/bot/data
- ./bot/.env:/srv/bot/.env:ro
environment:
PROCUREMENT_BOT_BOT_TOKEN: ${BALE_BOT_TOKEN:-1354444845:yhVMnKZBFuKaU3UW5fP1bHmknEHx99mipWw}
PROCUREMENT_BOT_BALE_API_BASE: ${BALE_API_BASE:-https://tapi.bale.ai}
PROCUREMENT_BOT_API_BASE_URL: http://api:8000/api/v1
PROCUREMENT_BOT_DATABASE_URL: postgresql+asyncpg://postgres:${POSTGRES_PASSWORD:-postgres}@postgres:5432/procurement
PROCUREMENT_BOT_AI_BASE_URL: ${AI_BASE_URL:-https://api.deepseek.com}
PROCUREMENT_BOT_AI_API_KEY: ${AI_API_KEY:-sk-a671eab15be846d494bf0839070a3b32}
PROCUREMENT_BOT_LLM_MODEL: ${AI_LLM_MODEL:-deepseek-flash}
PROCUREMENT_BOT_ADMIN_CHAT_IDS: ${BALE_ADMIN_CHAT_IDS:-1461166742,1839839922}
PROCUREMENT_BOT_SERVICE_TOKEN: ${BALE_SERVICE_TOKEN:-}
PROCUREMENT_BOT_FRONTEND_URL: ${APP_HOST:-http://localhost}
# Second copy of bot media in the ERP MinIO (empty user/key = off)
PROCUREMENT_BOT_MINIO_ENDPOINT: ${BOT_MINIO_ENDPOINT:-minio:9000}
PROCUREMENT_BOT_MINIO_ACCESS_KEY: ${MINIO_ROOT_USER:-minioadmin}
PROCUREMENT_BOT_MINIO_SECRET_KEY: ${MINIO_ROOT_PASSWORD:-minioadmin}
PROCUREMENT_BOT_MINIO_BUCKET: ${BOT_MINIO_BUCKET:-bot-media}
PROCUREMENT_BOT_MINIO_SECURE: "false"
depends_on:
postgres:
condition: service_healthy
api:
condition: service_started
networks:
- default
restart: unless-stopped
mem_limit: 512m
networks:
# شبکه‌ی عمومی: فقط ترافیک و سرویس‌های عمومی. postgres روی شبکه‌ی
# پیش‌فرض می‌ماند و از بیرون قابل دسترسی نیست.
traefik_network:
name: traefik_network
driver: bridge
# Fixed-name network shared with the monitoring stack (monitoring compose
# joins it as external) — carries /metrics scrapes and postgres-exporter.
shared:
name: erp_shared
driver: bridge
volumes:
postgres_data:
minio_data:
traefik_certs: