[Feat] 벡터 검색 인프라 구축 - OpenSQL+pgvector 및 bge-m3 임베딩 서버 - #38
Merged
Conversation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
📝 WalkthroughWalkthroughOpenSQL PostgreSQL에 pgvector 0.8.0 설치와 Changes데이터베이스 및 임베딩 인프라
Estimated code review effort: 4 (Complex) | ~45 minutes Sequence Diagram(s)sequenceDiagram
participant DockerCompose
participant PostgreSQL
participant EmbeddingServer
DockerCompose->>PostgreSQL: 로컬 OpenSQL 이미지 빌드 및 시작
PostgreSQL->>PostgreSQL: docgrid DB와 vector 확장 초기화
DockerCompose->>EmbeddingServer: embedding-server 빌드 및 시작
EmbeddingServer-->>DockerCompose: GET /health 상태 반환
Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
There was a problem hiding this comment.
Actionable comments posted: 4
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
docker-compose.yml (1)
17-22: 🩺 Stability & Availability | 🔴 Critical | ⚡ Quick winPostgreSQL의 Healthcheck가 실패하지 않도록 계정과 데이터베이스명을 업데이트하세요.
vars.yml파일 변경에 따라 기존app데이터베이스 및 사용자가 삭제되었습니다. 하지만 Healthcheck 명령은 여전히-U app -d app을 참조하고 있어 컨테이너가unhealthy상태에 빠지게 됩니다.경로 지시문(Path instructions)에 따른 환경 설정 검증 결과이며, 새롭게 생성된 DB 설정과 동기화가 필요합니다.
⚙️ Healthcheck 수정 제안
healthcheck: - test: ["CMD-SHELL", "/usr/pgsql-14/bin/pg_isready -U app -d app"] + test: ["CMD-SHELL", "/usr/pgsql-14/bin/pg_isready -U docgrid -d postgres"] interval: 10s timeout: 5s🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@docker-compose.yml` around lines 17 - 22, Update the PostgreSQL healthcheck command in the healthcheck configuration to use the database name and user defined by the updated vars.yml settings instead of the removed app account and database, keeping the existing readiness-check behavior unchanged.Source: Path instructions
🧹 Nitpick comments (1)
embedding-server/Dockerfile (1)
10-10: 🔒 Security & Privacy | 🔵 Trivial | ⚡ Quick win비특권(non-root) 사용자로 서버를 실행하도록 구성하는 것을 권장합니다.
운영 환경에서는 탈취된 컨테이너의 권한 남용을 방지하기 위해 컨테이너 내부 프로세스를
root가 아닌 일반 사용자로 실행하는 것이 보안 모범 사례입니다.🛡️ Non-root 사용자 추가 제안
COPY main.py . +RUN useradd -m appuser +USER appuser + CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@embedding-server/Dockerfile` at line 10, Update the embedding server Dockerfile to create and select a non-root user before the existing CMD entrypoint, ensuring uvicorn runs as that user while preserving the current host, port, and application settings.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@docker-compose.yml`:
- Around line 55-60: Update the healthcheck configuration’s start_period from
120s to at least 900s, allowing sufficient time for the embedding server’s
initial model download while leaving the existing test, interval, timeout, and
retries unchanged.
In `@docker/opensql/init-and-start.sh`:
- Around line 17-18: Update the pg_hba.conf entry appended by the initialization
script to use scram-sha-256 instead of trust, while preserving the existing
idempotent grep-and-append behavior. Ensure the configured database password is
used for remote authentication.
In `@embedding-server/main.py`:
- Line 44: Update the model.encode call in the embedding request flow to pass
normalize_embeddings=True, preserving the existing batch_size and max_length
options.
- Around line 12-19: Update lifespan so BGEM3FlagModel initialization runs in a
background thread or task instead of blocking startup, while preserving the
existing global model state. Allow the FastAPI server to begin accepting
requests immediately, keep model unset during loading so endpoints return the
existing 503 response, and retain cleanup by resetting model on shutdown.
---
Outside diff comments:
In `@docker-compose.yml`:
- Around line 17-22: Update the PostgreSQL healthcheck command in the
healthcheck configuration to use the database name and user defined by the
updated vars.yml settings instead of the removed app account and database,
keeping the existing readiness-check behavior unchanged.
---
Nitpick comments:
In `@embedding-server/Dockerfile`:
- Line 10: Update the embedding server Dockerfile to create and select a
non-root user before the existing CMD entrypoint, ensuring uvicorn runs as that
user while preserving the current host, port, and application settings.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: 9203303b-866b-46a7-820d-6824f6311b3b
📒 Files selected for processing (8)
docker-compose.ymldocker/opensql/Dockerfiledocker/opensql/init-and-start.shdocker/opensql/vars.ymldocs/kangcheolung-#35-embedding-server.mdembedding-server/Dockerfileembedding-server/main.pyembedding-server/requirements.txt
| def embed(req: EmbedRequest): | ||
| if model is None: | ||
| raise HTTPException(status_code=503, detail="Model not loaded") | ||
| result = model.encode([req.text], batch_size=1, max_length=8192) |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
PR 목표에 명시된 normalize_embeddings=True 적용이 누락되었습니다.
PR 요약 및 목표(PR objectives)에는 normalize_embeddings=True를 적용한다고 기재되어 있으나, model.encode() 호출에는 해당 옵션이 반영되어 있지 않습니다. 의도적으로 제외하신 것인지, 아니면 추가해야 하는지 확인 부탁드립니다.
💡 옵션 추가 제안
만약 라이브러리에서 해당 인자를 지원한다면 다음과 같이 추가할 수 있습니다:
- result = model.encode([req.text], batch_size=1, max_length=8192)
+ result = model.encode([req.text], batch_size=1, max_length=8192, normalize_embeddings=True)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| result = model.encode([req.text], batch_size=1, max_length=8192) | |
| result = model.encode([req.text], batch_size=1, max_length=8192, normalize_embeddings=True) |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@embedding-server/main.py` at line 44, Update the model.encode call in the
embedding request flow to pass normalize_embeddings=True, preserving the
existing batch_size and max_length options.
…m-sha-256 적용, 모델 백그라운드 로드 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This was referenced Jul 26, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
🔍️ 작업 내용
Closes #35
✨ 상세 설명
OpenSQL + pgvector 구성
tmaxopensql/postgres:14.6이미지에 pgvector 0.8.0을 소스 컴파일로 설치하는 커스텀 Dockerfile 추가init-and-start.sh: Ansible 완료 후 docgrid DB 자동 생성, vector 확장 활성화, pg_hba.conf TCP 접속 허용 설정임베딩 서버
POST /embed: 텍스트 입력 → 1024차원 dense vector 반환GET /health: 서버 및 모델 로드 상태 확인🛠 추후 리팩토링 및 고도화 계획
0.0.0.0/0 trust에서 Docker 브리지 대역으로 좁히기 (시연 환경 보안)POST /embed/batch)📸 스크린샷 (선택)
💬 리뷰 요구사항
docker/opensql/vars.yml변경으로 pg 유저가app → docgrid로 바뀌었습니다. 기존 볼륨이 있는 경우docker compose down -v후 재시작 필요합니다.docker compose up시 bge-m3 모델 다운로드로 약 10~15분 소요됩니다.Summary by CodeRabbit
새 기능
문서