Added traefik for production inference through tunnel

This commit is contained in:
Viswamedha Nalabotu 2026-03-21 00:05:00 +00:00
parent f58b4dbe35
commit 7becac6619

View file

@ -11,6 +11,7 @@ services:
- driver: nvidia - driver: nvidia
count: all count: all
capabilities: [gpu] capabilities: [gpu]
restart: unless-stopped
environment: environment:
- INFERENCE_HTTP_HOST=0.0.0.0 - INFERENCE_HTTP_HOST=0.0.0.0
- INFERENCE_HTTP_PORT=8001 - INFERENCE_HTTP_PORT=8001
@ -19,9 +20,19 @@ services:
- PYTHONPATH=/app - PYTHONPATH=/app
- HF_HOME=/root/.cache/huggingface - HF_HOME=/root/.cache/huggingface
- HF_HUB_OFFLINE=1 - HF_HUB_OFFLINE=1
ports:
- '0.0.0.0:58001:8001'
volumes: volumes:
- ../../:/app - ../../:/app
- ../../models:/app/models - ../../models:/app/models
- ../../hf_cache:/root/.cache/huggingface - ../../hf_cache:/root/.cache/huggingface
labels:
- "traefik.enable=true"
- "traefik.http.routers.llm.rule=Host(`${LLM_DOMAIN_NAME}`)"
- "traefik.http.routers.llm.entrypoints=web"
- "traefik.http.services.llm.loadbalancer.server.port=${LLM_PORT}"
- "traefik.docker.network=proxy-net"
networks:
- proxy-net
networks:
proxy-net:
external: true