Docker-compose, mysql, and the "Duplicate mount point" error. Help?

Docker Engine Compose
1

When I modify a mysql service in a docker-compose file, then try to apply the changes by running docker-compose stop/up/restart commands, I often have it throw the following error:

$ docker-compose up -d
Recreating mysql_1 ... 

Recreating mysql_1 ... error

ERROR: for mysql_1  Cannot create container for service mysql: Duplicate mount point: /var/lib/mysql

ERROR: for mysql  Cannot create container for service mysql: Duplicate mount point: /var/lib/mysql

I’ve been searching Google with several different searches trying to figure out what exactly is going on, but so far I haven’t hit on the magic words…

It seems like docker-compose is trying to create a new volume and use it when the old volume is still around. What I don’t understand is why it’s trying to create a new volume in the first place. The whole point of volumes is to keep the data around so it can used again…

Can anyone explain what’s going on?

My docker-compose.yml looks something like this:

version: "3.3"

volumes:
  mysql_data:

networks:
  appnet:
    driver: bridge
    driver_opts:
      com.docker.network.bridge.name: appnet

secrets:
  dbhost:
    file: ./secrets/dbhost
  dbname:
    file: ./secrets/dbname
  dbpass:
    file: ./secrets/dbpass
  dbport:
    file: ./secrets/dbport
  dbtype:
    file: ./secrets/dbtype
  dbuser:
    file: ./secrets/dbuser
  dbrootpass:
    file: ./secrets/dbrootpass
  selfurlpath:
    file: ./secrets/selfurlpath

services:
  app:
    image: registry.gitlab.com/jerrac/app/app_app:latest
    container_name: app_app
    ports:
      - 127.0.0.1:3050:80
    networks:
      - appnet
    depends_on:
      - app_mysql
    restart: always
    secrets:
      - dbhost
      - dbname
      - dbpass
      - dbport
      - dbtype
      - dbuser
      - selfurlpath
  app_mysql:
    image: mysql:5
    container_name: db_app
    volumes:
      - type: volume
        source: mysql_data
        target: /var/lib/mysql
      - type: bind
        source: ./datamount
        target: /srv
    environment:
      MYSQL_ROOT_PASSWORD_FILE: /run/secrets/dbrootpass
      MYSQL_DATABASE: app
      MYSQL_USER: app
      MYSQL_PASSWORD_FILE: /run/secrets/dbpass
    networks:
      - appnet
    restart: always
    secrets:
      - dbname
      - dbpass
      - dbuser
      - dbrootpass

I can get it to start again by deleting the old volume. But that’s hardly something I can do when I have real data in it… Oh, and I’ve run into this multiple times on several different projects.

Thanks!

Edit: Oops, forgot to add this info:

$ docker-compose version
docker-compose version 1.20.1, build 5d8c71b
docker-py version: 3.2.1
CPython version: 2.7.12
OpenSSL version: OpenSSL 1.0.2g  1 Mar 2016
$ cat /etc/lsb-release 
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=16.04
DISTRIB_CODENAME=xenial
DISTRIB_DESCRIPTION="Ubuntu 16.04.4 LTS"

Steps to reproduce:

  • Create docker-compose project with a mysql container that mounts a named volume.
  • docker up -d
  • Modify the mysql service.
  • docker-compose stop
  • docker-compose up -d
  • Or just docker-compose restart.

Issue type: Support request. I’m pretty sure I’m just not understanding how to do something correctly.

2

I just ran into this issue again. I still haven’t found a solid answer on what is actually going on. Any help would be much appreciated.

3

I was able to recreate this error as follows…

  1. In one terminal window fire-up a stack (docker-compose up) with a service that does not have a volume mapping declared. Let this run while we move to another terminal.
  2. In another window, edit the existing docker-compose.yml to add a volume mapping to that service
  3. In that second terminal run docker-compose up and the error appears

In my case, I had forgotten that the stack was already running when I started editing the config. Was pretty confusing, but that seems to be at the root of the error. Right thing to do is always shut down (docker-compose down) before editing the docker-compose.yml config.