Redis Cache By Xon

Redis Cache By Xon 2.18.6

No permission to download
@Xon, your site's download has 2.18.4 as the latest, even though 2.18.5 is selected. Just tested. XF.com's has 2.18.5. I kept downloading the last update from Atelier.
I had attached 2.18.4 instead of 2.18.5 for the XF2.3 version on my site. This has been fixed

The fix between 2.18.4 => 2.18.5 is;
  • Fix caching forum totals support when a date limit filter or 'Thread list date limit' is set on the forum
 
Given this add-on doesn't touch any search related functionality; yes that is very much unexpected and very likely not due to this add-on
 
If Redis is unavailable, will this add-on cause XF to gracefully fail back to a non-cached state, or will the site become unavailable?

My experience is that the stock Redis implementation causes the site to return 500 errors if the cache layer goes away, even if a timeout is set.
 
I set up a VDS with DirectAdmin, then immediately installed Redis and activated this plugin for XenForo. It’s truly a lifesaver—thank you again!
 
Given this add-on doesn't touch any search related functionality; yes that is very much unexpected and very likely not due to this add-on
I see. To be honest, when I set up the redis and enabled advanced search, the instant search wasn't working, but I didn't do a detailed investigation; I'm testing it in a test environment. I'll set up a test environment again and test it.
 
There are two separate XenForo forums installed on the same server. If the Redis add-on is active on both of them, will any problems occur?
 
set the namespace uniquely

 
Hello everyone,
Ekran görüntüsü 2025-11-15 151015.webpI have a server running AlmaLinux 10 + DirectAdmin + LiteSpeed.I’ve configured Redis to be used by both WordPress and XenForo, but I want to clarify a few details about the setup. Here’s an overview of my current configuration:

✔ REDIS SETUP ON THE SERVER (DirectAdmin User-Based Redis)​

There is no system-wide Redis installed on the server.
DirectAdmin creates a localized (user-based) Redis instance for each user.
Mine runs through this socket:

/home/admin/.redis/redis.sock

So I’m using a Unix socket instead of a port.
This Redis instance works without issues, and both XenForo and WordPress connect through the same socket.

✔ PHP SIDE (phpredis) STATUS​


To use Redis, the phpredis extension must be active on the PHP side.

  • PHP 8.4 has phpredis enabled → working without issues
  • LiteSpeed is also using PHP 8.4, so WordPress connects properly
  • XenForo shows “phpredis 6.3.0” in the Redis information panel
In other words, phpredis is active for both WordPress and XenForo.

✔ XENFORO CONFIGURATION​


I’m using the RedisCache add-on.
My config is:

$config['cache']['provider'] = 'SV\RedisCache\Redis';
$config['cache']['config'] = [
'server' => 'unix:///home/admin/.redis/redis.sock',
'port' => null,
];
$config['cache']['namespace'] = 'xf_';


$config['pageCache']['enabled'] = true;
$config['pageCache']['lifetime'] = 300;

The XenForo Redis Information page is fully populated:
  • Redis 8.2.3
  • phpredis 6.3.0
  • Memory / TTL / IOPS metrics look normal
So XenForo is fully using Redis.

✔ WORDPRESS CONFIGURATION​

I have LiteSpeed Cache installed, but I disabled LiteSpeed’s built-in Object Cache.
Instead, I’m using Redis Object Cache (drop-in).

  • Status: Connected
  • Drop-in: Valid
  • Metrics page shows activity
  • phpredis client active
  • Redis → Reachable

So WordPress is also fully working on Redis.

✔ WHAT I WANT TO CONFIRM​

Both WordPress and XenForo are using the same Redis socket and the same phpredis extension.

Current working structure:

  • Redis Server: DirectAdmin’s per-user Redis instance
  • Connection: Unix socket
  • PHP Client: phpredis
  • Caching:
    • WordPress → Redis Object Cache plugin
    • XenForo → SV RedisCache add-on
  • LiteSpeed Object Cache: Disabled (to prevent conflicts)

Is this architecture correct?
Would using the same Redis instance for both WordPress and XenForo cause any issues in the long run?


To avoid key collisions, I set namespace = xf_ in XenForo.
WordPress already uses its own prefixes.

If necessary, I can move XenForo to a separate Redis database (DB 1), but for now I don’t see any key conflicts.

✔ TL;DR​

  • Redis working ✔
  • phpredis active ✔
  • WordPress Redis Object Cache working ✔
  • XenForo Redis Cache working ✔
  • LiteSpeed Object Cache disabled ✔
  • Both apps using a shared Redis instance via Unix socket ✔

Is there anything in this setup that could cause conflicts, performance issues, or that goes against best practices?
I’d appreciate insights from users with experience.
 
Back
Top Bottom