---
tags:
  - project/sriphat
  - airflow
  - workflow
  - etl
created: 2026-05-07
status: active
folder: 05-airflow
---

# Apache Airflow (05-airflow)

> **Docker Compose:** `05-airflow/docker-compose.yaml`
> **Env File:** `05-airflow/.env`
> **Version:** Apache Airflow 3.1.5

## Overview

Apache Airflow ใช้สำหรับ Workflow Orchestration:
- รัน DAGs (Directed Acyclic Graphs) แบบตั้งเวลา
- ประมวลผล Excel/CSV files จาก Finance
- ETL pipeline orchestration
- Integration กับ API Service

**Executor:** CeleryExecutor (ใช้ Redis เป็น broker)

---

## Services

| Container | หน้าที่ | Port |
|-----------|--------|------|
| `airflow-apiserver` | REST API + Web UI | `8200:8080` |
| `airflow-scheduler` | DAG scheduling | internal |
| `airflow-dag-processor` | DAG file parsing | internal |
| `airflow-worker` | Task execution (Celery) | internal |
| `airflow-triggerer` | Deferred task triggering | internal |
| `airflow-init` | Database migration (one-time) | — |
| `airflow-cli` | CLI tool (debug profile) | — |
| `flower` | Celery monitoring (optional) | `5555:5555` |

---

## Architecture

```
                 ┌─────────────────┐
                 │  airflow-        │
                 │  apiserver       │  ← Web UI + REST API (port 8200)
                 │  (port 8080)     │
                 └────────┬────────┘
                          │
          ┌───────────────┼───────────────┐
          │               │               │
   ┌──────▼──────┐ ┌──────▼──────┐ ┌─────▼──────┐
   │  airflow-   │ │  airflow-   │ │  airflow-  │
   │  scheduler  │ │  dag-       │ │  triggerer │
   │             │ │  processor  │ │            │
   └──────┬──────┘ └─────────────┘ └────────────┘
          │
          ▼  (Celery tasks via Redis)
   ┌──────────────┐
   │  airflow-    │
   │  worker      │  ← รัน tasks จริง
   └──────────────┘
          │
          ▼
   ┌──────────────┐
   │  PostgreSQL  │  (Airflow metadata DB)
   │  Redis       │  (Celery broker)
   └──────────────┘
```

---

## Database Configuration

Airflow ใช้ PostgreSQL บน Infra server:

```bash
# Connection string
AIRFLOW__DATABASE__SQL_ALCHEMY_CONN=
  postgresql+psycopg2://${AIRFLOW_DB_USER}:${AIRFLOW_DB_PASSWD}@${AIRFLOW_DB_HOST}:${AIRFLOW_DB_PORT}/${AIRFLOW_DB_NAME}

AIRFLOW__CELERY__RESULT_BACKEND=
  db+postgresql://${AIRFLOW_DB_USER}:${AIRFLOW_DB_PASSWD}@${AIRFLOW_DB_HOST}:${AIRFLOW_DB_PORT}/${AIRFLOW_DB_NAME}

# Redis broker
AIRFLOW__CELERY__BROKER_URL=redis://:@redis:6379/0
```

---

## Volume Mounts

```
05-airflow/
├── dags/        → /opt/airflow/dags      (DAG files)
├── logs/        → /opt/airflow/logs      (Task logs)
├── config/      → /opt/airflow/config    (airflow.cfg)
│   └── airflow.cfg
└── plugins/     → /opt/airflow/plugins   (Custom plugins)
```

---

## Web UI

**URL:** `http://localhost:8200` หรือ `https://ai.sriphat.com/airflow`

```bash
# Config
AIRFLOW__WEBSERVER__BASE_URL=https://ai.sriphat.com/airflow
AIRFLOW__WEBSERVER__WEB_SERVER_PORT=8080
```

Default credentials (ถ้าไม่เปลี่ยน):
- Username: `airflow`
- Password: `airflow`

---

## DAGs ที่มีอยู่

| DAG ID | หน้าที่ | ถูก Trigger จาก |
|--------|--------|----------------|
| `process_finance_excel` | ประมวลผล Excel ของ Finance | API Service |

---

## Airflow Configuration (airflow.cfg)

**Path:** `05-airflow/config/airflow.cfg`

Key settings:
```ini
[core]
executor = CeleryExecutor
load_examples = False
dags_are_paused_at_creation = True

[webserver]
base_url = https://ai.sriphat.com/airflow

[execution_api]
execution_api_server_url = http://airflow-apiserver:8080/execution/
```

---

## Environment Variables

```bash
# Airflow image
AIRFLOW_IMAGE_NAME=apache/airflow:3.1.5

# Database
AIRFLOW_DB_USER=<user>
AIRFLOW_DB_PASSWD=<password>
AIRFLOW_DB_HOST=<postgres-host>
AIRFLOW_DB_PORT=5432
AIRFLOW_DB_NAME=airflow

# Security
AIRFLOW__CORE__FERNET_KEY=<fernet-key>

# Admin user
_AIRFLOW_WWW_USER_USERNAME=airflow
_AIRFLOW_WWW_USER_PASSWORD=<password>

# Optional pip packages
_PIP_ADDITIONAL_REQUIREMENTS=
```

---

## Deploy Commands

```bash
cd 05-airflow

# Initialize (first time only)
docker compose up airflow-init

# Start all services
docker compose up -d

# View logs
docker logs airflow-apiserver -f
docker logs airflow-scheduler -f
docker logs airflow-worker -f

# Run Celery Flower monitoring
docker compose --profile flower up -d

# Scale workers (เพิ่ม worker)
docker compose up -d --scale airflow-worker=3
```

---

## System Requirements

Airflow ต้องการ resources ขั้นต่ำ:
- **RAM:** ≥ 4 GB
- **CPU:** ≥ 2 cores
- **Disk:** ≥ 10 GB

---

## Ingestion Layer (04-ingestion / Airbyte)

> **หมายเหตุ:** `04-ingestion/docker-compose.yml` ปัจจุบัน **commented out ทั้งหมด**
> Airbyte ถูก deploy แยกต่างหาก (ผ่าน `abctl` หรือ standalone)

### Airbyte ที่ระบุในแผน

| Source | ชนิดข้อมูล |
|--------|----------|
| SQL Server (HIS) | ข้อมูลผู้ป่วย, OPD |
| Oracle (Lab) | ผลตรวจทางห้องปฏิบัติการ |
| REST API | External data |
| Excel/CSV | Finance, รายงาน |

**Destination:** PostgreSQL `raw_data` schema

**Port:** `8030` (เมื่อ deploy แล้ว)

---

## Related

- [[00-Project-Overview]]
- [[01-Infrastructure]]
- [[03-API-Service]]
- [[08-Operations-Runbook]]