postgresql保留数据：更好地命名卷或绑定安装

Question

Option 1: (named container. the volume is identified by its name. It store its data in the /var/lib/docker/volumes/nameofthevolume)

# create the volume in advance
  $ docker volume create test_vol

Option: 2 (here name of the volume bind-test does not matter, what matter is which local path /home/user/test it mounts to, which is persistant. Rather than /var/lib/docker/volume/somevolumename /home/user/somedatafolder makes more readability. Cons: we have to ensure that the /home/user/somedatafolder exists.)

 # inside a docker-compose file
  ...
  volumes:
    bind-test:
      driver: local
      driver_opts:
        type: none
        o: bind
        device: /home/user/test

 or:

version: '3'
services:
myservice:
    volumes:
      - ./path:/volume/path

The downside of bind mounts is that it places files that are managed by containers, with the uid/gid from the container, inside a path likely used by other users on the host, often with a different uid/gid on the host. The result is permission issues either on the host or inside the container. You need to align uid/gid's between the two to avoid this.

Answer 1

At the end of the day, there isn't a big difference between bind mount and Docker named volumes.

I tend to prefer keeping persistent data from Docker services in Docker volumes. You can then use tools like docker system df -v to inspect what your application uses.

As for exporting the data, you can use docker cp

docker cp someContainer:/somedir/ .