Sebastian Krüger
f668e06228
All checks were successful
Build and Push RunPod Docker Image / build-and-push (push) Successful in 36s
- Create config_bge.yaml for BAAI/bge-large-en-v1.5 on port 8002 - Reduce Llama VRAM to 70% and context to 16K for concurrent use - Add BGE service to supervisor with vllm group 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
144 lines
3.6 KiB
Plaintext
144 lines
3.6 KiB
Plaintext
[supervisord]
|
|
logfile=.logs/supervisord.log
|
|
pidfile=supervisord.pid
|
|
childlogdir=.logs
|
|
nodaemon=false
|
|
loglevel=info
|
|
|
|
[unix_http_server]
|
|
file=supervisor.sock
|
|
chmod=0700
|
|
|
|
[supervisorctl]
|
|
serverurl=unix://supervisor.sock
|
|
|
|
[rpcinterface:supervisor]
|
|
supervisor.rpcinterface_factory = supervisor.rpcinterface:make_main_rpcinterface
|
|
|
|
# Web interface for monitoring (proxied via nginx + Authelia)
|
|
[inet_http_server]
|
|
port=0.0.0.0:9001
|
|
# Authentication disabled - handled by Authelia SSO
|
|
|
|
# ComfyUI Server
|
|
[program:comfyui]
|
|
command=services/comfyui/venv/bin/python services/comfyui/main.py --listen 0.0.0.0 --port 8188 --enable-cors-header --preview-method auto
|
|
directory=.
|
|
autostart=true
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/comfyui.err.log
|
|
stdout_logfile=.logs/comfyui.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=PYTORCH_CUDA_ALLOC_CONF="expandable_segments:True",TQDM_DISABLE="1"
|
|
priority=100
|
|
stopwaitsecs=30
|
|
|
|
|
|
# ComfyUI WebDAV Sync Service
|
|
[program:webdav-sync]
|
|
command=services/webdav-sync/venv/bin/python services/webdav-sync/webdav_sync.py
|
|
directory=.
|
|
autostart=true
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/webdav-sync.err.log
|
|
stdout_logfile=.logs/webdav-sync.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=WEBDAV_URL="%(ENV_WEBDAV_URL)s",WEBDAV_USERNAME="%(ENV_WEBDAV_USERNAME)s",WEBDAV_PASSWORD="%(ENV_WEBDAV_PASSWORD)s",WEBDAV_REMOTE_PATH="%(ENV_WEBDAV_REMOTE_PATH)s",COMFYUI_OUTPUT_DIR="./services/comfyui/output"
|
|
priority=150
|
|
stopwaitsecs=10
|
|
|
|
|
|
# vLLM Llama 3.1 8B Server (Port 8001)
|
|
[program:llama]
|
|
command=services/vllm/venv/bin/vllm serve --config services/vllm/config_llama.yaml
|
|
directory=.
|
|
autostart=false
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/llama.err.log
|
|
stdout_logfile=.logs/llama.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=HF_HOME="./.cache/vllm",HF_TOKEN="%(ENV_HF_TOKEN)s"
|
|
priority=200
|
|
stopwaitsecs=30
|
|
|
|
|
|
# vLLM BGE Embedding Server (Port 8002)
|
|
[program:bge]
|
|
command=services/vllm/venv/bin/vllm serve --config services/vllm/config_bge.yaml
|
|
directory=.
|
|
autostart=false
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/bge.err.log
|
|
stdout_logfile=.logs/bge.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=HF_HOME="./.cache/vllm",HF_TOKEN="%(ENV_HF_TOKEN)s"
|
|
priority=210
|
|
stopwaitsecs=30
|
|
|
|
|
|
# AudioCraft Studio Service
|
|
[program:audiocraft]
|
|
command=services/audiocraft/venv/bin/python services/audiocraft/main.py
|
|
directory=.
|
|
autostart=false
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/audiocraft.err.log
|
|
stdout_logfile=.logs/audiocraft.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=HF_HOME="./.cache/audiocraft",HF_TOKEN="%(ENV_HF_TOKEN)s",AUDIOCRAFT_ROOT_PATH="https://audiocraft.ai.pivoine.art"
|
|
priority=300
|
|
stopwaitsecs=10
|
|
|
|
# Upscale UI Service
|
|
[program:upscale]
|
|
command=services/upscale/venv/bin/python services/upscale/app.py
|
|
directory=.
|
|
autostart=false
|
|
autorestart=true
|
|
startretries=3
|
|
stderr_logfile=.logs/upscale.err.log
|
|
stdout_logfile=.logs/upscale.out.log
|
|
stdout_logfile_maxbytes=50MB
|
|
stdout_logfile_backups=10
|
|
stderr_logfile_maxbytes=50MB
|
|
stderr_logfile_backups=10
|
|
environment=UPSCALE_PORT="8080"
|
|
priority=400
|
|
stopwaitsecs=10
|
|
|
|
[group:comfyui]
|
|
programs=comfyui,webdav-sync
|
|
priority=100
|
|
|
|
[group:vllm]
|
|
programs=llama,bge
|
|
priority=200
|
|
|
|
[group:audiocraft]
|
|
programs=audiocraft
|
|
priority=300
|
|
|
|
[group:upscale]
|
|
programs=upscale
|
|
priority=400
|