도커 컴포즈를 사용한 PGVector 환경 구축

개요

PostgreSQL에 PGVector 확장을 추가해서 랭체인 벡터 저장소로 활용한다.

디렉토리 구성

Directoryvector-search/
- docker-compose.yaml
- Directorypostgres/
  - schema.sql

PGVector 도커 컴포즈 설정

작업 디렉토리 생성 및 docker-compose.yaml 파일 생성

mkdir vector-search
cd vector-search
touch docker-compose.yaml
mkdir postgres
cd postgres
touch schema.sql

schema.sql 파일 편집

-- Enable pgvector extension
CREATE EXTENSION IF NOT EXISTS vector;

-- Create sample table
CREATE TABLE items (
    id SERIAL PRIMARY KEY,
    name VARCHAR(255) NOT NULL,
    item_data JSONB,
    embedding vector(1536) -- vector data
);

docker-compose.yaml 파일 편집

services:
  pgvector-db:
    image: pgvector/pgvector:pg17
    container_name: pgvector-db
    environment:
      POSTGRES_USER: langchain
      POSTGRES_PASSWORD: langchain
      POSTGRES_DB: langchain
    ports:
      - "5432:5432"
    volumes:
      - pgdata:/var/lib/postgresql/data
      - ./postgres/schema.sql:/docker-entrypoint-initdb.d/schema.sql

volumes:
  pgdata: # Stores data outside the container to ensure persistence

PostgreSQL 기동
Terminal window
```
docker-compose up -d
```

PostgreSQL 연결

docker exec -it pgvector-db psql -U langchain -d langchain

확장 목록 조회

-- Check installed extensions
\dx

  Name   | Version |   Schema   |                     Description
---------+---------+------------+------------------------------------------------------
 plpgsql | 1.0     | pg_catalog | PL/pgSQL procedural language
 vector  | 0.8.1   | public     | vector data type and ivfflat and hnsw access methods
(2 rows)

테이블 생성 확인

-- Check table creation
\dt

 Schema | Name  | Type  |   Owner
--------+-------+-------+-----------
 public | items | table | langchain
(1 row)

items 테이블 구성 조회

-- Check table structure
\d items

  Column   |          Type          | Collation | Nullable |              Default
-----------+------------------------+-----------+----------+-----------------------------------
 id        | integer                |           | not null | nextval('items_id_seq'::regclass)
 name      | character varying(255) |           | not null |
 item_data | jsonb                  |           |          |
 embedding | vector(1536)           |           |          |
Indexes:
    "items_pkey" PRIMARY KEY, btree (id)

참조문서 링크

Build Your Own AI Search Engine (3 Part Series)