1. Build multi-platform images

1.1 Build & Push by GitHub Actions

You can use the GitHub Actions workflow or any other CI tools, to build and push multi-platform images to Docker container registry. Refer to the .github/workflows/build-and-push.yml file for more details. Trigger build action on push of tags starting with “v”, e.g., v1.0.0

./deploy/Dockerfile is a multi-stage Dockerfile with three stages:

fonts — downloads Noto font files via deploy/install-font.sh (cached layer, rarely changes)
build_stage — compiles the application jar with Maven
final image — copies the fonts and jar into a minimal JRE image

The APP_PATH, APP_NAME parameters are passed as build arguments in the GitHub Actions workflow. The image version is extracted from the tag name e.g., v1.0.0 -> 1.0.0.

./deploy/Dockerfile.ci is a two-stage variant for CI pipelines where the jar is pre-built. It includes the same font installation stage.

1.2 Build & Push by Local Script Manually

Build images and push them to Docker container registry. Specify the APP_PATH, APP_VERSION to build the image for the Java application.


./deploy/build.sh <APP_PATH> <APP_VERSION> [<REGISTRY_NAMESPACE>]

The APP_PATH is the relative path to the application source code directory, such as apps/demo-app. And the last name in APP_PATH is the application name, such as demo-app.

The REGISTRY_NAMESPACE is optional, and the default value is softa. You can specify the REGISTRY_NAMESPACE to push the image to your own docker image repository.

Example to build the demo application image:


./deploy/build.sh apps/demo-app 1.0.3

2. Install Noto Fonts (PDF Generation)

Noto fonts are required for PDF generation with file-starter (RICH_TEXT document templates). The fonts are installed automatically inside Docker images (see above). For local development, run:


sh deploy/install-font.sh

The script auto-detects the OS font directory (~/Library/Fonts on macOS, ~/.local/share/fonts on Linux). You can also pass an explicit target directory:


sh deploy/install-font.sh /path/to/fonts

3. Start EFK by Docker Compose (Optional)


docker-compose -f deploy/efk/docker-compose.yml up -d

Access the Kibana console at http://localhost:5601

Or you can specify the spring.elasticsearch.uris property to connect to your own Elasticsearch cluster.

Then create index using demo-app/init_es/create_index.

4. Start Pulsar by Docker Compose (Optional)


docker-compose -f deploy/pulsar/docker-compose.yml up -d

Access the Pulsar console at http://localhost:8080

Or you can specify the spring.pulsar.client.service-url property to connect to your own Pulsar cluster.

The most crucial information is you need to configure the mq.topics.xxx.topic properties to enable the corresponding Listeners.

On the other hand, if you are not ready to set up the pulsar service, you can also choose not to configure or comment out the mq.topics.xxx topics to avoid the issue of being unable to start.

5. Start Minio by Docker Compose (Optional)


docker-compose -f deploy/minio/docker-compose.yml up -d

Minio API Endpoints

http://localhost:9000

Minio Web UI Dashboard

http://localhost:9001 Username: minioadmin Password: minioadmin

6. Start the demo application by Docker Compose


docker-compose -f ./deploy/demo-app/docker-compose.yml up -d

Create a database instance and execute the SQL scripts in deploy/demo-app/init_mysql.

7. Production Environment

It is highly recommended that the production environment be deployed using a pipeline and Kubernetes for containerization.