Configure SQL Server 2017 container images on Docker
This article explains how to configure and use the mssql-server-linux container image with Docker. This image consists of SQL Server running on Linux based on Ubuntu 16.04. It can be used with the Docker Engine 1.8+ on Linux or on Docker for Mac/Windows.
This article specifically focuses on using the mssql-server-linux image. The Windows image is not covered, but you can learn more about it on the mssql-server-windows Docker Hub page.
Pull and run the container image
To pull and run the Docker container image for SQL Server 2017, follow the prerequisites and steps in the following quickstart:
This configuration article provides additional usage scenarios in the following sections.
Run production container images
The quickstart in the previous section runs the free Developer edition of SQL Server from Docker Hub. Most of the information still applies if you want to run production container images, such as Enterprise, Standard, or Web editions. However, there are a few differences that are outlined here.
You can only use SQL Server in a production environment if you have a valid license. You can obtain a free SQL Server Express production license here. SQL Server Standard and Enterprise Edition licenses are available through Microsoft Volume Licensing.
Production SQL Server container images must be pulled from Docker Store. If you don't already have one, create an account on Docker Store.
The Developer container image on Docker Store can be configured to run the production editions as well. Use the following steps to run production editions:
First, log in to your docker id from the command line.
Next, you need to obtain the free Developer container image on Docker Store. Go to https://store.docker.com/images/mssql-server-linux, click Proceed to Checkout, and follow the instructions.
Review the requirements and run procedures in the quickstart. But there are two differences. You must pull the image store/microsoft/mssql-server-linux:<tag-name> from Docker Store. And you must specify your production edition with the MSSQL_PID environment variable. The following example shows how to run the latest SQL Server 2017 container image for the Enterprise Edition:
docker run --name sqlenterprise \ -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' \ -e 'MSSQL_PID=Enterprise' -p 1433:1433 \ -d store/microsoft/mssql-server-linux:2017-latest
docker run --name sqlenterprise ` -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" ` -e "MSSQL_PID=Enterprise" -p 1433:1433 ` -d "store/microsoft/mssql-server-linux:2017-latest"
By passing the value Y to the environment variable ACCEPT_EULA and an edition value to MSSQL_PID, you are expressing that you have a valid and existing license for the edition and version of SQL Server that you intend to use. You also agree that your use of SQL Server software running in a Docker container image will be governed by the terms of your SQL Server license.
For a full list of possible values for MSSQL_PID, see Configure SQL Server settings with environment variables on Linux.
Connect and query
You can connect and query SQL Server in a container from either outside the container or from within the container. The following sections explain both scenarios.
Tools outside the container
You can connect to the SQL Server instance on your Docker machine from any external Linux, Windows, or macOS tool that supports SQL connections. Some common tools include:
The following example uses sqlcmd to connect to SQL Server running in a Docker container. The IP address in the connection string is the IP address of the host machine that is running the container.
sqlcmd -S 10.3.2.4 -U SA -P '<YourPassword>'
sqlcmd -S 10.3.2.4 -U SA -P "<YourPassword>"
If you mapped a host port that was not the default 1433, add that port to the connection string. For example, if you specified
-p 1400:1433 in your
docker run command, then connect by explicitly specify port 1400.
sqlcmd -S 10.3.2.4,1400 -U SA -P '<YourPassword>'
sqlcmd -S 10.3.2.4,1400 -U SA -P "<YourPassword>"
Tools inside the container
Starting with SQL Server 2017 CTP 2.0, the SQL Server command-line tools are included in the container image. If you attach to the image with an interactive command-prompt, you can run the tools locally.
docker exec -itcommand to start an interactive bash shell inside your running container. In the following example
e69e056c702dis the container ID.
docker exec -it e69e056c702d "bash"
You don't always have to specify the entire container id. You only have to specify enough characters to uniquely identify it. So in this example, it might be enough to use
e69rather than the full id.
Once inside the container, connect locally with sqlcmd. Note that sqlcmd is not in the path by default, so you have to specify the full path.
/opt/mssql-tools/bin/sqlcmd -S localhost -U SA -P '<YourPassword>'
When finished with sqlcmd, type
When finished with the interactive command-prompt, type
exit. Your container continues to run after you exit the interactive bash shell.
Run multiple SQL Server containers
Docker provides a way to run multiple SQL Server containers on the same host machine. This is the approach for scenarios that require multiple instances of SQL Server on the same host. Each container must expose itself on a different port.
The following example creates two SQL Server containers and maps them to ports 1401 and 1402 on the host machine.
docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1401:1433 -d microsoft/mssql-server-linux:2017-latest docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1402:1433 -d microsoft/mssql-server-linux:2017-latest
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1401:1433 -d microsoft/mssql-server-linux:2017-latest docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1402:1433 -d microsoft/mssql-server-linux:2017-latest
Now there are two instances of SQL Server running in separate containers. Clients can connect to each SQL Server instance by using the IP address of the Docker host and the port number for the container.
sqlcmd -S 10.3.2.4,1401 -U SA -P '<YourPassword>' sqlcmd -S 10.3.2.4,1402 -U SA -P '<YourPassword>'
sqlcmd -S 10.3.2.4,1401 -U SA -P "<YourPassword>" sqlcmd -S 10.3.2.4,1402 -U SA -P "<YourPassword>"
Persist your data
Your SQL Server configuration changes and database files are persisted in the container even if you restart the container with
docker stop and
docker start. However, if you remove the container with
docker rm, everything in the container is deleted, including SQL Server and your databases. The following section explains how to use data volumes to persist your database files even if the associated containers are deleted.
For SQL Server, it is critical that you understand data persistence in Docker. In addition to the discussion in this section, see Docker's documentation on how to manage data in Docker containers.
Mount a host directory as data volume
The first option is to mount a directory on your host as a data volume in your container. To do that, use the
docker run command with the
-v <host directory>:/var/opt/mssql flag. This allows the data to be restored between container executions.
docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1433:1433 -v <host directory>:/var/opt/mssql -d microsoft/mssql-server-linux:2017-latest
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1433:1433 -v <host directory>:/var/opt/mssql -d microsoft/mssql-server-linux:2017-latest
This technique also enables you to share and view the files on the host outside of Docker.
Host volume mapping for Docker on Mac with the SQL Server on Linux image is not supported at this time. Use data volume containers instead. This restriction is specific to the
/var/opt/mssql directory. Reading from a mounted directory works fine. For example, you can mount a host directory using –v on Mac and restore a backup from a .bak file that resides on the host.
Use data volume containers
The second option is to use a data volume container. You can create a data volume container by specifying a volume name instead of a host directory with the
-v parameter. The following example creates a shared data volume named sqlvolume.
docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1433:1433 -v sqlvolume:/var/opt/mssql -d microsoft/mssql-server-linux:2017-latest
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1433:1433 -v sqlvolume:/var/opt/mssql -d microsoft/mssql-server-linux:2017-latest
This technique for implicitly creating a data volume in the run command does not work with older versions of Docker. In that case, use the explicit steps outlined in the Docker documentation, Creating and mounting a data volume container.
Even if you stop and remove this container, the data volume persists. You can view it with the
docker volume ls command.
docker volume ls
If you then create another container with the same volume name, the new container uses the same SQL Server data contained in the volume.
To remove a data volume container, use the
docker volume rm command.
If you delete the data volume container, any SQL Server data in the container is permanently deleted.
Backup and restore
In addition to these container techniques, you can also use standard SQL Server backup and restore techniques. You can use backup files to protect your data or to move the data to another SQL Server instance. For more information, see Backup and restore SQL Server databases on Linux.
If you do create backups, make sure to create or copy the backup files outside of the container. Otherwise, if the container is removed, the backup files are also deleted.
Execute commands in a container
If you have a running container, you can execute commands within the container from a host terminal.
To get the container ID run:
To start a bash terminal in the container run:
docker exec -ti <Container ID> /bin/bash
Now you can run commands as though you are running them at the terminal inside the container. When finished, type
exit. This exits in the interactive command session, but your container continues to run.
Copy files from a container
To copy a file out of the container, use the following command:
docker cp <Container ID>:<Container path> <host path>
docker cp d6b75213ef80:/var/opt/mssql/log/errorlog /tmp/errorlog
docker cp d6b75213ef80:/var/opt/mssql/log/errorlog C:\Temp\errorlog
Copy files into a container
To copy a file into the container, use the following command:
docker cp <Host path> <Container ID>:<Container path>
docker cp /tmp/mydb.mdf d6b75213ef80:/var/opt/mssql/data
docker cp C:\Temp\mydb.mdf d6b75213ef80:/var/opt/mssql/data
Run a specific SQL Server container image
There are scenarios where you might not want to use the latest SQL Server container image. To run a specific SQL Server container image, use the following steps:
Identify the Docker tag for the release you want to use. To view the available tags, see the mssql-server-linux Docker hub page.
Pull the SQL Server container image with the tag. For example, to pull the RC1 image, replace
<image_tag>in the following command with
docker pull microsoft/mssql-server-linux:<image_tag>
To run a new container with that image, specify the tag name in the
docker runcommand. In the following command, replace
<image_tag>with the version you want to run.
docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1401:1433 -d microsoft/mssql-server-linux:<image_tag>
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1401:1433 -d microsoft/mssql-server-linux:<image_tag>
These steps can also be used to downgrade an existing container. For example, you might want to rollback or downgrade a running container for troubleshooting or testing. To downgrade a running container, you must be using a persistence technique for the data folder. Follow the same steps outlined in the upgrade section, but specify the tag name of the older version when you run the new container.
Upgrade and downgrade are only supported between RC1 and RC2 at this time.
Upgrade SQL Server in containers
To upgrade the container image with Docker, first identify the tag for the release for your upgrade. Pull this version from the registry with the
docker pull command:
docker pull microsoft/mssql-server-linux:<image_tag>
This updates the SQL Server image for any new containers you create, but it does not update SQL Server in any running containers. To do this, you must create a new container with the latest SQL Server container image and migrate your data to that new container.
Make sure you are using one of the data persistence techniques for your existing SQL Server container. This enables you to start a new container with the same data.
Stop the SQL Server container with the
Create a new SQL Server container with
docker runand specify either a mapped host directory or a data volume container. Make sure to use the specific tag for the your SQL Server upgrade. The new container now uses a new version of SQL Server with your existing SQL Server data.
Upgrade is only supported between RC1, RC2, and GA at this time.
Verify your databases and data in the new container.
Optionally, remove the old container with
The following sections provide troubleshooting suggestions for running SQL Server in containers.
Docker command errors
If you get errors for any
docker commands, make sure that the docker service is running, and try to run with elevated permissions.
For example, on Linux, you might get the following error when running
Cannot connect to the Docker daemon. Is the docker daemon running on this host?
If you get this error on Linux, try running the same commands prefaced with
sudo. If that fails, verify the docker service is running, and start it if necessary.
sudo systemctl status docker sudo systemctl start docker
On Windows, verify that you are launching PowerShell or your command-prompt as an Administrator.
SQL Server container startup errors
If the SQL Server container fails to run, try the following tests:
If you get an error such as 'failed to create endpoint CONTAINER_NAME on network bridge. Error starting proxy: listen tcp 0.0.0.0:1433 bind: address already in use.', then you are attempting to map the container port 1433 to a port that is already in use. This can happen if you're running SQL Server locally on the host machine. It can also happen if you start two SQL Server containers and try to map them both to the same host port. If this happens, use the
-pparameter to map the container port 1433 to a different host port. For example:
docker run -e 'ACCEPT_EULA=Y' -e 'MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>' -p 1400:1433 -d microsoft/mssql-server-linux:2017-latest`.
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -p 1400:1433 -d microsoft/mssql-server-linux:2017-latest`.
Check to see if there are any error messages from container.
docker logs e69e056c702d
Make sure that you meet the minimum memory and disk requirements specified in the Requirements section of this topic.
If you are using any container management software, make sure it supports container processes running as root. The sqlservr process in the container runs as root.
Review the SQL Server setup and error logs.
Enable dump captures
If the SQL Server process is failing inside the container, you should create a new container with SYS_PTRACE enabled. This adds the Linux capability to trace a process, which is necessary for creating a dump file on an exception. The dump file can be used by support to help troubleshoot the problem. The following docker run command enables this capability.
docker run -e "ACCEPT_EULA=Y" -e "MSSQL_SA_PASSWORD=<YourStrong!Passw0rd>" -e "MSSQL_PID=Developer" --cap-add SYS_PTRACE -p 1401:1433 -d microsoft/mssql-server-linux:2017-latest
SQL Server connection failures
If you can't connect to the SQL Server instance running in your container, try the following tests:
Make sure that your SQL Server container is running by looking at the STATUS column of the
docker ps -aoutput. If not, use
docker start <Container ID>to start it.
If you mapped to a non-default host port (not 1433), make sure you are specifying the port in your connection string. You can see your port mapping in the PORTS column of the
docker ps -aoutput. For example, the following command connects sqlcmd to a container listening on port 1401:
sqlcmd -S 10.3.2.4,1401 -U SA -P '<YourPassword>'
sqlcmd -S 10.3.2.4,1401 -U SA -P "<YourPassword>"
If you used
docker runwith an existing mapped data volume or data volume container, SQL Server ignores the value of
MSSQL_SA_PASSWORD. Instead, the pre-configured SA user password is used from the SQL Server data in the data volume or data volume container. Verify that you are using the SA password associated with the data you're attaching to.
Review the SQL Server setup and error logs.
SQL Server Availability Groups
If you are using Docker with SQL Server Availability Groups, there are two additional requirements.
Map the port that is used for replica communication (default 5022). For example, specify
-p 5022:5022as part of your
Explicitly set the container host name with the
-h YOURHOSTNAMEparameter of the
docker runcommand. This host name is used when you configure your Availability Group. If you don't specify it with
-h, it defaults to the container ID.
SQL Server setup and error logs
You can look at the SQL Server setup and error logs in /var/opt/mssql/log. If the container is not running, first start the container. Then use an interactive command-prompt to inspect the logs.
docker start e69e056c702d docker exec -it e69e056c702d "bash"
From the bash session inside your container, run the following commands:
cd /var/opt/mssql/log cat setup*.log cat errorlog
If you mounted a host directory to /var/opt/mssql when you created your container, you can instead look in the log subdirectory on the mapped path on the host.
Get started with SQL Server 2017 container images on Docker by going through the quickstart.
Also, see the mssql-docker GitHub repository for resources, feedback, and known issues.