Vaessl/docs/02-Preparation/02-Spring-Boot-and-database-setup.md

Vaessl: Spring Boot and database setup

This app will use the current latest version 4.0.4 of Spring Boot and the latest OpenJDK 25 LTS.

The dependencies are chosen to specifically work with Spring AI and PostgreSQL/pgvector. For the AI function OpenAI is chosen since it is known to work with LiteLLM. The PostgreSQL dependency makes sure to include PostGreSQL support for self hosted databases.

Spring Initializr settings

Build System: Gradle - Kotlin

Gradel Kotlin is chosen to make mobile/Android app development easier.

Spring Boot: 4.0.4

Packaging: Jar

Configuration: YAML

Java: 25

Dependencies

• Lombok
• Spring Boot DevTools
• Spring Web
• Spring Security
• Spring Data JPA
• OpenAI AI
• PostgreSQL Driver
• Validation

Project Settings

PostGreSQL and OpenAI need an initial setup so that the local instance is able to start. I will comment out Spring Security since user management is an issue for a later iteration of the app.

The build.gradle.kts will look something like this:

plugins {
	java
	id("org.springframework.boot") version "4.0.4"
	id("io.spring.dependency-management") version "1.1.7"
}

group = "com.vaessl"
version = "0.0.1-SNAPSHOT"

java {
	toolchain {
		languageVersion = JavaLanguageVersion.of(25)
	}
}

configurations {
	compileOnly {
		extendsFrom(configurations.annotationProcessor.get())
	}
}

repositories {
	mavenCentral()
}

extra["springAiVersion"] = "2.0.0-M3"

dependencies {
	implementation("org.springframework.boot:spring-boot-starter-data-jpa")
	// implementation("org.springframework.boot:spring-boot-starter-security")
	implementation("org.springframework.boot:spring-boot-starter-validation")
	implementation("org.springframework.boot:spring-boot-starter-webmvc")
	implementation("org.springframework.ai:spring-ai-starter-model-openai")
	compileOnly("org.projectlombok:lombok")
	developmentOnly("org.springframework.boot:spring-boot-devtools")
	runtimeOnly("org.postgresql:postgresql")
	annotationProcessor("org.projectlombok:lombok")
	testImplementation("org.springframework.boot:spring-boot-starter-data-jpa-test")
	// testImplementation("org.springframework.boot:spring-boot-starter-security-test")
	testImplementation("org.springframework.boot:spring-boot-starter-validation-test")
	testImplementation("org.springframework.boot:spring-boot-starter-webmvc-test")
	testRuntimeOnly("org.junit.platform:junit-platform-launcher")
	testImplementation("org.springframework.boot:spring-boot-starter-test")
}

dependencyManagement {
	imports {
		mavenBom("org.springframework.ai:spring-ai-bom:${property("springAiVersion")}")
	}
}

tasks.withType<Test> {
	useJUnitPlatform()
}

To configure OpenAI (which I will use instead of LiteLLM initially since I have an OpenAI Api key and can get going quickly) and PostGreSQL I will create a .env.local file in the root dir, fill all credentials and add it to .gitignore

DB_URL=jdbc:postgresql://192.168.1.208:5433/vaessl
DB_TEST_URL=jdbc:postgresql://192.168.1.208:5434/vaessl_test
DB_USERNAME=myusername
DB_PASSWORD=mypw
OPENAI_KEY=myapikey
OPENAI_BASE_URL=https://api.openai.com
PG_DRIVER_CLASS_NAME=org.postgresql.Driver

The initial application.yaml in the resources folder will look like this:

spring:
  application:
    name: vaessl
  config:
    import: "optional:file:.env.local[.properties]"
  datasource:
    url : ${DB_URL}
    username: ${DB_USERNAME}
    password: ${DB_PASSWORD}
    driver-class-name: ${PG_DRIVER_CLASS_NAME}

  jpa:
    hibernate:
      ddl-auto: update
    show-sql: true

  ai:
    openai:
      base-url: ${OPENAI_BASE_URL}
      api-key: ${OPENAI_KEY}
      chat:
        options:
          model: gpt-4o-mini

  logging:
    level:
      org.springframework.boot.context.config: TRACE

The Docker Compose file for code-server will look something like this:

---
services:
  code-server:
    image: code-server-dev:latest
    container_name: code-server
    environment:
      - PUID=1000
      - PGID=1000
      - TZ=Europe/Vienna
#      - PASSWORD= #optional
      - HASHED_PASSWORD=hashedpw
#      - SUDO_PASSWORD_HASH= #optional
      - PROXY_DOMAIN=code-server.your.website
      - DEFAULT_WORKSPACE=/config/workspace
    volumes:
      - /home/pi/docker/vscode:/config
      - /home/pi/docker/vscode/workspace:/workspace
    ports:
      - 8443:8443
      - 8124:8080
      - 5173:5173
    restart: unless-stopped

Note that I'm using my own locally hosted PostgreSQL instances for the main and test database. Just add databases via SQL or PgAdmin and install the pgvector extension to each database manually. There is an official ready-made pgvector docker image but if you already host a PostgreSQL database you need to add the extension yourself.

Installing pgvector on a self-hosted PostgreSQL container

pgvector is a PostgreSQL extension that adds a vector data type. It does not create a new database — it adds a vector_store table to your existing database. The shared library (vector.so) must be installed on the PostgreSQL server itself.

Do not install it manually inside a running container. Manual installations do not survive container recreation (e.g. after a docker compose up --force-recreate or image update). Instead, bake it into a custom Docker image.

Step 1: Create a custom Dockerfile for PostgreSQL

Create a Dockerfile.postgres next to your docker-compose.yaml:

FROM postgres:18.4

RUN apt-get update && apt-get install -y --no-install-recommends \
    ca-certificates git build-essential postgresql-server-dev-18 \
    && git clone --branch v0.8.3 https://github.com/pgvector/pgvector.git \
    && cd pgvector && make && make install \
    && cd .. && rm -rf pgvector \
    && apt-get purge -y ca-certificates git build-essential postgresql-server-dev-18 \
    && apt-get autoremove -y \
    && rm -rf /var/lib/apt/lists/*

• FROM postgres:18.4 — this is still the standard postgres image, not the pgvector image
• ca-certificates — required for git to verify GitHub's SSL certificate during build
• postgresql-server-dev-18 — provides the PostgreSQL header files needed to compile pgvector

Step 2: Update docker-compose.yaml to use the custom image

services:
  db:
    container_name: postgres
    labels:
      - "com.centurylinklabs.watchtower.enable=false"
    build:
      context: .
      dockerfile: Dockerfile.postgres
      network: host
    restart: always
    environment:
      POSTGRES_USER: ${DB_USER}
      POSTGRES_PASSWORD: ${DB_PASSWORD}
      POSTGRES_DB: ${DB_NAME}
    networks:
      - pg_network
    volumes:
      - /home/pi/docker/postgresql:/var/lib/postgresql
    ports:
      - "5432:5432"

networks:
  pg_network:
    external: true

Step 3: Build the image and recreate the container

Before recreating, back up your databases:

docker exec -it <container-name> bash
su - postgres -c "pg_dumpall > /tmp/backup.sql"
exit
docker cp <container-name>:/tmp/backup.sql .

Build the image. Use --network=host to ensure the build container can reach GitHub:

docker build --network=host -t postgres-pgvector -f Dockerfile.postgres .

Recreate the container

docker compose up -d --force-recreate

Step 4: Enable the extensions in each database

Run this once per database that needs pgvector:

docker exec -it <container-name> psql -U <db-user> -d <db-name>

CREATE EXTENSION IF NOT EXISTS vector;
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";

uuid-ossp is also required — Spring AI's vector_store table uses uuid_generate_v4() for its primary key.

Step 5: Add pgvector config to application.yaml

Spring AI's pgvector auto-configuration requires these properties:

spring:
  ai:
    vectorstore:
      pgvector:
        dimensions: 1536          # must match your embedding model output size
        distance-type: COSINE_DISTANCE
        index-type: HNSW
        initialize-schema: true   # auto-creates the vector_store table on startup

dimensions depends on the embedding model:

• text-embedding-ada-002 / text-embedding-3-small → 1536
• text-embedding-3-large → 3072

initialize-schema: true means Spring AI will create the vector_store table automatically on first startup — no manual SQL needed.

Appendix: Additional config for developing in Code-Server

When using the code-server container there are additional config steps to mind:

• Assign the 8080 port to a different port if it is used by another docker container by adding a port variable in the docker-compose.yaml. I assigned it to 8124.

  ports:
  	- 8443:8443
  	- 8124:8080
  

• define a proxy domain for automatic Url generations when starting localhost

  environment:
  	- PROXY_DOMAIN=code-server.your.website
  

  This makes sure port 8080 is reachable via https://8080.code-server.your.website as per code-server documentation for using subdomains: https://coder.com/docs/code-server/guide#using-a-subdomain

• make sure to add the subdomain in your proxy platform like Cloudflare Zero Trust Tunnel or Pangolin and point it to the local ip in my case http://192.168.208:8124