Redis connection down? Stop the “Error Establishing a Redis Connection” crash on WordPress

If you’re running the Redis Object Cache plugin on a client site, you’ve probably seen this ugly white-screen error at some point:

Error establishing a Redis connection

To disable Redis, delete the object-cache.php file in the /wp-content/ directory.

By default, if Redis becomes unreachable (server restart, memory limit hit, network hiccup, provider maintenance..) the plugin throws a fatal error and takes your entire site down with it. Not great for a caching layer that’s supposed to make things faster, not riskier.

The One-Line Fix

Add this to wp-config.php:

define( 'WP_REDIS_GRACEFUL', true );

That’s it. With this constant set, if Redis goes down, WordPress doesn’t throw a fatal error – it just silently stops using Redis for that request and falls back to WordPress’s built-in in-memory (non-persistent) cache instead. Your site stays up. Visitors never see an error page. You get a clean entry in your PHP error log instead of an outage.

---

How does it actually work under the hood

Every time a Redis operation fails, the plugin calls an internal handle_exception() method. Without WP_REDIS_GRACEFUL, that method dies with wp_die() and shows the error screen above. With it enabled, the plugin instead:

• Marks Redis as disconnected for the current request
• Falls back all cache groups to the non-persistent, in-request cache
• Logs the exception quietly via error_log()
• Lets the page render normally

Worth knowing: this fallback resets on every request. It’s not a sticky “Redis is down, stop trying for 5 minutes” circuit breaker: WordPress will attempt to reconnect on the next page load too. If Redis is down for an extended period, every request pays a small connection-timeout cost before falling back. Keeping WP_REDIS_TIMEOUT and WP_REDIS_READ_TIMEOUT low (the plugin defaults to 1 second) keeps that cost negligible.

My standard practice

I add WP_REDIS_GRACEFUL to every client site where we set up Redis object caching. It costs nothing, and it’s the difference between “Redis had a bad five minutes” and “the site was down for five minutes.” If you want to get alerted when it kicks in rather than just checking logs, hook into redis_object_cache_error and fire off a Slack or email notification.

TL;DR: one constant, zero downside, add it every time.