ADR 008: Adopt fallback mechanism

[bmf-san]

Here’s the translation for the new request:

---

Title

Adopt Fallback Mechanism

Context

In a reverse proxy, maintaining a good user experience when upstream servers are unavailable or failures occur is important. A fallback mechanism is a critical feature that improves system fault tolerance and enhances the user experience.

Decision

It has been decided to implement a fallback mechanism with the following features:

• Upstream failure detection: Early detection of failures through health checks.
• Multiple fallback options:
  • Forwarding to alternative upstream servers
  • Serving static content
  • Displaying custom error pages
• Flexible configuration: YAML-based settings to customize fallback behavior for each path or host.

proxy:
  fallback:
    enabled: true
    health_check:
      interval: "10s"
      timeout: "5s"
      healthy_threshold: 2
      unhealthy_threshold: 3
    
    upstreams:
      - host: "primary.example.com"
        fallback_to:
          - "backup1.example.com"
          - "backup2.example.com"
        static_fallback: "/fallback"
        cache_ttl: "1h"
    
    paths:
      - path: "/api"
        fallback_to: "api-backup.example.com"
      - path: "/static"
        static_fallback: "/local-static"

Status

Accepted

Consequences

Positive

• Increased system availability
• Maintained user experience
• Flexible response to failures
• Reduced operational costs

Negative

• Increased complexity in configuration
• Need for testing fallback logic
• Additional storage requirements (for static fallback content)