Adolfo Delorenzo
ee2f88e13b
- Create llm_pool/router.py: LiteLLM Router with fast (Ollama) and quality (Anthropic/OpenAI) model groups - Configure fallback chain: quality providers fail -> fast group - Pin LiteLLM to ==1.82.5 (avoid September 2025 OOM regression in later releases) - Create llm_pool/main.py: FastAPI service on port 8004 with /complete and /health endpoints - Add providers/__init__.py: reserved for future per-provider customization - Update docker-compose.yml: add llm-pool and celery-worker service stubs
136 lines
3.6 KiB
YAML
136 lines
3.6 KiB
YAML
version: "3.9"
|
|
|
|
networks:
|
|
konstruct-net:
|
|
driver: bridge
|
|
|
|
volumes:
|
|
postgres_data:
|
|
redis_data:
|
|
ollama_data:
|
|
|
|
services:
|
|
postgres:
|
|
image: postgres:16-alpine
|
|
container_name: konstruct-postgres
|
|
environment:
|
|
POSTGRES_DB: konstruct
|
|
POSTGRES_USER: postgres
|
|
POSTGRES_PASSWORD: postgres_dev
|
|
volumes:
|
|
- postgres_data:/var/lib/postgresql/data
|
|
- ./scripts/init-db.sh:/docker-entrypoint-initdb.d/init-db.sh:ro
|
|
ports:
|
|
- "5432:5432"
|
|
networks:
|
|
- konstruct-net
|
|
healthcheck:
|
|
test: ["CMD-SHELL", "pg_isready -U postgres -d konstruct"]
|
|
interval: 5s
|
|
timeout: 5s
|
|
retries: 10
|
|
|
|
redis:
|
|
image: redis:7-alpine
|
|
container_name: konstruct-redis
|
|
command: redis-server --save 60 1 --loglevel warning
|
|
volumes:
|
|
- redis_data:/data
|
|
ports:
|
|
- "6379:6379"
|
|
networks:
|
|
- konstruct-net
|
|
healthcheck:
|
|
test: ["CMD", "redis-cli", "ping"]
|
|
interval: 5s
|
|
timeout: 5s
|
|
retries: 10
|
|
|
|
ollama:
|
|
image: ollama/ollama:latest
|
|
container_name: konstruct-ollama
|
|
volumes:
|
|
- ollama_data:/root/.ollama
|
|
ports:
|
|
- "11434:11434"
|
|
networks:
|
|
- konstruct-net
|
|
deploy:
|
|
resources:
|
|
reservations:
|
|
devices:
|
|
- driver: nvidia
|
|
count: all
|
|
capabilities: [gpu]
|
|
# Service starts even if no GPU is available — GPU config is optional
|
|
restart: unless-stopped
|
|
|
|
llm-pool:
|
|
build:
|
|
context: .
|
|
dockerfile_inline: |
|
|
FROM python:3.12-slim
|
|
WORKDIR /app
|
|
RUN pip install uv
|
|
COPY pyproject.toml ./
|
|
COPY packages/shared ./packages/shared
|
|
COPY packages/llm-pool ./packages/llm-pool
|
|
RUN uv pip install --system -e packages/shared -e packages/llm-pool
|
|
CMD ["uvicorn", "llm_pool.main:app", "--host", "0.0.0.0", "--port", "8004"]
|
|
container_name: konstruct-llm-pool
|
|
ports:
|
|
- "8004:8004"
|
|
networks:
|
|
- konstruct-net
|
|
depends_on:
|
|
ollama:
|
|
condition: service_started
|
|
redis:
|
|
condition: service_healthy
|
|
environment:
|
|
- ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY:-}
|
|
- OPENAI_API_KEY=${OPENAI_API_KEY:-}
|
|
- OLLAMA_BASE_URL=http://ollama:11434
|
|
- REDIS_URL=redis://redis:6379/0
|
|
- LOG_LEVEL=INFO
|
|
restart: unless-stopped
|
|
healthcheck:
|
|
test: ["CMD-SHELL", "curl -sf http://localhost:8004/health || exit 1"]
|
|
interval: 10s
|
|
timeout: 5s
|
|
retries: 5
|
|
|
|
celery-worker:
|
|
build:
|
|
context: .
|
|
dockerfile_inline: |
|
|
FROM python:3.12-slim
|
|
WORKDIR /app
|
|
RUN pip install uv
|
|
COPY pyproject.toml ./
|
|
COPY packages/shared ./packages/shared
|
|
COPY packages/orchestrator ./packages/orchestrator
|
|
RUN uv pip install --system -e packages/shared -e packages/orchestrator
|
|
CMD ["celery", "-A", "orchestrator.main", "worker", "--loglevel=info"]
|
|
container_name: konstruct-celery-worker
|
|
networks:
|
|
- konstruct-net
|
|
depends_on:
|
|
redis:
|
|
condition: service_healthy
|
|
postgres:
|
|
condition: service_healthy
|
|
llm-pool:
|
|
condition: service_healthy
|
|
environment:
|
|
- DATABASE_URL=postgresql+asyncpg://konstruct_app:konstruct_dev@postgres:5432/konstruct
|
|
- REDIS_URL=redis://redis:6379/0
|
|
- CELERY_BROKER_URL=redis://redis:6379/1
|
|
- CELERY_RESULT_BACKEND=redis://redis:6379/2
|
|
- LLM_POOL_URL=http://llm-pool:8004
|
|
- ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY:-}
|
|
- OPENAI_API_KEY=${OPENAI_API_KEY:-}
|
|
- OLLAMA_BASE_URL=http://ollama:11434
|
|
- LOG_LEVEL=INFO
|
|
restart: unless-stopped
|