Docker部署Apache Airflow时PostgreSQL pg_wal权限报错求助

尝试用Docker部署Apache Airflow时，Postgres容器启动异常，核心错误为：

LOG:  could not link file "pg_wal/xlogtemp.47" to "pg_wal/000000010000000000000001": Permission denied
FATAL:  could not open file "pg_wal/000000010000000000000001": No such file or directory

完整容器日志如下：

doc-postgres-1           | selecting default max_connections ... 100
doc-postgres-1           | selecting default shared_buffers ... 128MB
doc-postgres-1           | selecting default time zone ... Etc/UTC
doc-postgres-1           | creating configuration files ... ok
doc-postgres-1           | running bootstrap script ... 2022-11-17 08:18:14.665 UTC [47] LOG:  could not link file "pg_wal/xlogtemp.47" to "pg_wal/000000010000000000000001": Permission denied
doc-postgres-1           | 2022-11-17 08:18:14.674 UTC [47] FATAL:  could not open file "pg_wal/000000010000000000000001": No such file or directory
doc-postgres-1           | child process exited with exit code 1
doc-postgres-1           | initdb: removing contents of data directory "/var/lib/postgresql/data"
doc-postgres-1 exited with code 4294967295
doc-postgres-1           | The files belonging to this database system will be owned by user "postgres".
doc-postgres-1           | This user must also own the server process.
doc-postgres-1           |
doc-postgres-1           | The database cluster will be initialized with locale "en_US.utf8".
doc-postgres-1           | The default database encoding has accordingly been set to "UTF8".
doc-postgres-1           | The default text search configuration will be set to "english".
doc-postgres-1           |
doc-postgres-1           | Data page checksums are disabled.
doc-postgres-1           |
doc-postgres-1           | fixing permissions on existing directory /var/lib/postgresql/data ... ok
doc-postgres-1           | creating subdirectories ... ok
doc-postgres-1           | selecting dynamic shared memory implementation ... posix
doc-postgres-1           | selecting default max_connections ... 100
doc-postgres-1           | selecting default shared_buffers ... 128MB
doc-postgres-1           | selecting default time zone ... Etc/UTC
doc-postgres-1           | creating configuration files ... ok
doc-postgres-1           | running bootstrap script ... 2022-11-17 08:18:25.475 UTC [42] LOG:  could not link file "pg_wal/xlogtemp.42" to "pg_wal/000000010000000000000001": Permission denied
doc-postgres-1           | 2022-11-17 08:18:25.489 UTC [42] FATAL:  could not open file "pg_wal/000000010000000000000001": No such file or directory
doc-postgres-1           | child process exited with exit code 1
doc-postgres-1           | initdb: removing contents of data directory "/var/lib/postgresql/data"
container for service "postgres" is unhealthy

解决方案 #

1. 修复本地挂载目录权限 #

如果使用本地目录挂载Postgres数据卷，先清理旧数据并设置正确权限：

# 停止并删除所有相关容器和卷
docker-compose down -v
# 删除本地挂载的Postgres数据目录（替换为你的实际路径）
sudo rm -rf ./postgres-data
# 重新创建目录并赋予Postgres容器用户权限（UID=999是Postgres镜像默认用户ID）
mkdir -p ./postgres-data
sudo chown -R 999:999 ./postgres-data
# 重新启动容器
docker-compose up -d

2. 使用Docker命名卷替代本地目录 #

本地文件系统可能不支持硬链接（如Windows共享目录、NFS挂载），改用Docker命名卷可避开此类问题：
修改docker-compose.yml中Postgres的volumes配置：

services:
  postgres:
    image: postgres:13  # 对应Airflow兼容的Postgres版本
    volumes:
      - postgres_data:/var/lib/postgresql/data
    environment:
      POSTGRES_USER: airflow
      POSTGRES_PASSWORD: airflow
      POSTGRES_DB: airflow

# 在文件末尾添加命名卷定义
volumes:
  postgres_data:

然后重新启动容器：docker-compose up -d

3. 临时禁用本地挂载（测试环境） #

若仅用于测试，可直接去掉本地卷挂载，让容器使用内部存储：

services:
  postgres:
    image: postgres:13
    # 注释掉本地挂载行
    # volumes:
    #   - ./postgres-data:/var/lib/postgresql/data
    environment:
      POSTGRES_USER: airflow
      POSTGRES_PASSWORD: airflow
      POSTGRES_DB: airflow

4. 检查Docker运行权限 #

确保当前用户已加入docker组，或使用sudo执行docker-compose命令，避免因权限不足导致容器无法操作挂载目录：

# 将当前用户加入docker组
sudo usermod -aG docker $USER
# 重新登录后生效，或直接用sudo启动
sudo docker-compose up -d

内容的提问来源于stack exchange，提问作者laxman dangi