The Optimized MySQL Installation for Docker

Docker images for MySQL are optimized for code size, which means they only include crucial components that are expected to be relevant for the majority of users who run MySQL instances in Docker containers. A MySQL Docker installation is different from a common, non-Docker installation in the following aspects:

Configuring the MySQL Server

When you start the MySQL Docker container, you can pass configuration options to the server through the docker run command. For example:

docker run --name mysql1 -d mysql/mysql-server:tag --character-set-server=utf8mb4 --collation-server=utf8mb4_col

The command starts your MySQL Server with utf8mb4 as the default character set and utf8mb4_col as the default collation for your databases.

Another way to configure the MySQL Server is to prepare a configuration file and mount it at the location of the server configuration file inside the container. See Persisting Data and Configuration Changes for details.

Persisting Data and Configuration Changes

Docker containers are in principle ephemeral, and any data or configuration are expected to be lost if the container is deleted or corrupted (see discussions here). Docker volumes, however, provides a mechanism to persist data created inside a Docker container. At its initialization, the MySQL Server container creates a Docker volume for the server data directory. The JSON output for running the docker inspect command on the container has a Mount key, whose value provides information on the data directory volume:

$> docker inspect mysql1
...
 "Mounts": [
            {
                "Type": "volume",
                "Name": "4f2d463cfc4bdd4baebcb098c97d7da3337195ed2c6572bc0b89f7e845d27652",
                "Source": "/var/lib/docker/volumes/4f2d463cfc4bdd4baebcb098c97d7da3337195ed2c6572bc0b89f7e845d27652/_data",
                "Destination": "/var/lib/mysql",
                "Driver": "local",
                "Mode": "",
                "RW": true,
                "Propagation": ""
            }
        ],
...

The output shows that the source folder /var/lib/docker/volumes/4f2d463cfc4bdd4baebcb098c97d7da3337195ed2c6572bc0b89f7e845d27652/_data, in which data is persisted on the host, has been mounted at /var/lib/mysql, the server data directory inside the container.

Another way to preserve data is to bind-mount a host directory using the --mount option when creating the container. The same technique can be used to persist the configuration of the server. The following command creates a MySQL Server container and bind-mounts both the data directory and the server configuration file:

docker run --name=mysql1 \
--mount type=bind,src=/path-on-host-machine/my.cnf,dst=/etc/my.cnf \
--mount type=bind,src=/path-on-host-machine/datadir,dst=/var/lib/mysql \
-d mysql/mysql-server:tag 

The command mounts path-on-host-machine/my.cnf at /etc/my.cnf (the server configuration file inside the container), and path-on-host-machine/datadir at /var/lib/mysql (the data directory inside the container). The following conditions must be met for the bind-mounting to work:

Running Additional Initialization Scripts

If there are any .sh or .sql scripts you want to run on the database immediately after it has been created, you can put them into a host directory and then mount the directory at /docker-entrypoint-initdb.d/ inside the container. For example:

docker run --name=mysql1 \
--mount type=bind,src=/path-on-host-machine/scripts/,dst=/docker-entrypoint-initdb.d/ \
-d mysql/mysql-server:tag

Connect to MySQL from an Application in Another Docker Container

By setting up a Docker network, you can allow multiple Docker containers to communicate with each other, so that a client application in another Docker container can access the MySQL Server in the server container. First, create a Docker network:

docker network create my-custom-net

Then, when you are creating and starting the server and the client containers, use the --network option to put them on network you created. For example:

docker run --name=mysql1 --network=my-custom-net -d mysql/mysql-server

docker run --name=myapp1 --network=my-custom-net -d myapp

The myapp1 container can then connect to the mysql1 container with the mysql1 hostname and vice versa, as Docker automatically sets up a DNS for the given container names. In the following example, we run the mysql client from inside the myapp1 container to connect to host mysql1 in its own container:

docker exec -it myapp1 mysql --host=mysql1 --user=myuser --password

For other networking techniques for containers, see the Docker container networking section in the Docker Documentation.

Server Error Log

When the MySQL Server is first started with your server container, a server error log is NOT generated if either of the following conditions is true:

To make MySQL Server generate an error log when either of the two conditions is true, use the --log-error option to configure the server to generate the error log at a specific location inside the container. To persist the error log, mount a host file at the location of the error log inside the container as explained in Persisting Data and Configuration Changes. However, you must make sure your MySQL Server inside its container has write access to the mounted host file.

Known Issues

Docker Environment Variables

When you create a MySQL Server container, you can configure the MySQL instance by using the --env option (-e in short) and specifying one or more of the following environment variables.