- Complete Fastify gateway with 8-stage pipeline - Circuit breaker (opossum) per model tier - Rate limiting per caller - Ban list validation (EN/DE/auto-detected) - TIP validator (SFF-8024, part numbers, wavelengths) - Prometheus metrics - pg-boss async queue - PostgreSQL audit log + review queue - 9 prompt templates (TIP, LinkedIn, ShieldX) - Learning engine scaffolding - Auto-learning: ban-list, few-shot, routing, prompt optimizer
54 lines
1.5 KiB
Nginx Configuration File
54 lines
1.5 KiB
Nginx Configuration File
# Nginx reverse proxy for LLM Gateway
|
|
# Place at: /etc/nginx/sites-available/llm-gateway
|
|
# Enable: ln -s /etc/nginx/sites-available/llm-gateway /etc/nginx/sites-enabled/
|
|
# Reload: nginx -t && systemctl reload nginx
|
|
#
|
|
# NOTE: If using Cloudflare Tunnel, nginx is optional.
|
|
# The tunnel connects directly to localhost:3100.
|
|
|
|
upstream llm_gateway {
|
|
server localhost:3100;
|
|
keepalive 32;
|
|
}
|
|
|
|
server {
|
|
listen 80;
|
|
server_name llm-gateway.context-x.org;
|
|
|
|
# Security headers
|
|
add_header X-Content-Type-Options nosniff;
|
|
add_header X-Frame-Options DENY;
|
|
add_header X-XSS-Protection "1; mode=block";
|
|
|
|
# Don't leak nginx version
|
|
server_tokens off;
|
|
|
|
location / {
|
|
proxy_pass http://llm_gateway;
|
|
|
|
# Timeouts: 130s to handle large LLM responses without proxy interruption
|
|
proxy_read_timeout 130s;
|
|
proxy_connect_timeout 10s;
|
|
proxy_send_timeout 130s;
|
|
|
|
# Standard proxy headers
|
|
proxy_set_header Host $host;
|
|
proxy_set_header X-Real-IP $remote_addr;
|
|
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
|
|
proxy_set_header X-Forwarded-Proto $scheme;
|
|
|
|
# Keep-alive to upstream
|
|
proxy_http_version 1.1;
|
|
proxy_set_header Connection "";
|
|
|
|
# Buffer streaming responses (LLMs can stream tokens)
|
|
proxy_buffering off;
|
|
}
|
|
|
|
# Health endpoint — no logging (noisy in monitoring)
|
|
location /health {
|
|
proxy_pass http://llm_gateway/health;
|
|
access_log off;
|
|
}
|
|
}
|