Built-in Redis cache does not gracefully degrade on connection failure, leading to fatal errors

[bzcomputers]

Affected version

2.0+

When the built-in Redis cache provider is configured and Redis becomes temporarily unavailable (e.g., during a service restart or brief outage), XenForo throws an uncaught RedisException, resulting in a fatal error that makes the entire site inaccessible. This occurs early in the bootstrap process during DataRegistry operations, preventing the site from loading any content until Redis recovers.

Expected: The system should degrade gracefully by falling back to a non-Redis cache (e.g., file-based or null) for the duration of the outage, keeping the site at least operational, albeit slower.

Expected Behavior​

• Catch Redis connection exceptions in early cache reads (e.g., DataRegistry).
• Fallback to a built-in alternative cache (e.g., Filesystem or Null) for the request.
• Log the exception without halting execution.
• Keep the site accessible during short Redis outages.

Actual Behavior​

• Uncaught RedisException propagates from the provider.
• Fatal error halts bootstrap — full site downtime until Redis is restored.
• Nothing logged within XenForo.

Suggested Fix​

• Wrap cache calls with try/catch for RedisException/CacheException.
• On catch, switch to fallback provider (e.g., new \XF\Cache\FilesystemAdapter).
• Log: \XF::logException($e, false);
• Add config option: $config['cache']['failure_fallback'] = 'Filesystem'; (default: none/throw).