Migrate via Docker Container

Streamline your migration execution with Docker images with yuniql built-in. Use Docker CLI to run your migrations in any Docker compatible host.

Run your database migration thru a Docker container. This is specially helpful on Linux environments and CI/CD pipelines running on Linux Agents as it facilitates your migration without having to worry any local installations or runtime dependencies.

When you run yuniql init command, a baseline directory structure will be created automatically. This includes a ready-to-use Dockerfile.

When you call docker build, we pull the base image containing the nightly build of yuniql and all of your local structure is copied into the image. When you call docker run, it executes a yuniql run inside the container and targets your desired server.

NOTE: The container must have access to the target database and server. You may need to configure a firewall rule to accept login requests from the container hosts especially for cloud-based databases.

Pre-requisites

Prepare database migration project

Install yuniql CLI. Install yuniql CLI with Chocolatey or use alternative ways listed here https://yuniql.io/docs/install-yuniql/.

choco install yuniql --version=1.0.1

Create new migration workspace.

md c:\temp\yuniql-docker
cd c:\temp\yuniql-docker

yuniql init
dir /O:N

10/21/2019  22:41    <DIR>          _draft
10/21/2019  22:41    <DIR>          _erase
10/21/2019  22:41    <DIR>          _init
10/21/2019  22:41    <DIR>          _post
10/21/2019  22:41    <DIR>          _pre
10/21/2019  22:41    <DIR>          v0.00
10/21/2019  22:41                   Dockerfile
10/21/2019  22:41                   README.md
10/21/2019  22:41                   .gitignore

Create first script file setup_tables.sql on v0.00.

CREATE TABLE [dbo].[Visitor](
	[VisitorID] [int] IDENTITY(1000,1) NOT NULL,
	[FirstName] [nvarchar](255) NULL,
	[LastName] [varchar](255) NULL,
	[Address] [nvarchar](255) NULL,
	[Email] [nvarchar](255) NULL
) ON [PRIMARY];
GO

Build docker image.

docker build -t sqlserver-example .

Run migration from docker.

docker run sqlserver-example -c "<your-connection-string>" -a
docker logs sqlserver-example -f

Commit your project into git and use it as input in creating CI/CD pipelines.

Setup Azure DevOps Pipelines

The following pipelines runs sqlserver-sample project into Azure SQL Database. The samples are available on Yuniql GitHub repository. You may clone the repo to test the pipelines.

Agent task: docker build
steps:
- task: [email protected]2
  displayName: 'docker build'
  inputs:
    command: build
    Dockerfile: '$(System.DefaultWorkingDirectory)/_rdagumampan_yuniql/samples/basic-sqlserver-sample/Dockerfile'
    arguments: '-t sqlserver-example'
    addPipelineData: false
Agent task: docker run
variables:
  AzSqlDemoDatabase: '<YOUR-SQLDATABASE-CONNECTIONSTRING>'

steps:
- task: [email protected]2
  displayName: 'docker run'
  inputs:
    command: run
    arguments: 'sqlserver-example -c "$(AzSqlDemoDatabase)" --debug'
    addPipelineData: false

Private Container Registry

If your company security policy disallows pulling images DockerHub, you may build the base image internally by cloning the GitHub repo and building docker images locally. The docker files are multi-staged. Please refer to dockerfile.multi-stage-linux-x64 and dockerfile.multi-stage-win-x64.

Build images and push to your preferred registry

docker build -t yuniql -f dockerfile.multi-stage-linux-x64 .
docker tag yuniql:latest rdagumampan/yuniql:linux-x64-latest

docker login -u="%DOCKERHUB_USERNAME%" -p="%DOCKERHUB_PASSWORD%"
docker push yuniql:linux-x64-latest
docker build -t yuniql -f dockerfile.multi-stage-win-x64 .
docker tag yuniql:latest rdagumampan/yuniql:win-x64-latest

docker login -u="%DOCKERHUB_USERNAME%" -p="%DOCKERHUB_PASSWORD%"
docker push yuniql:win-x64-latest

Modify the Docker file of your database project

#FROM rdagumampan/yuniql:linux-x64-latest
FROM <your-internal-repository>/yuniql:linux-x64-latest
COPY . ./db

Learn further

Found bugs?

Help us improve further please create an issue.

Comments