AccelerateWP’s Object Cache now runs on Valkey, the open-source fork of Redis. Every site that uses it switches over on its own when the server takes the update, so on most servers no site needs a change. AccelerateWP 1.11-1 brings the change and is now in beta. Stable follows after a gradual rollout. We’ll update this post when it gets there.
Until now, Object Cache gave each hosting user their own Redis 6.2 instance. Redis 6.2 reaches its upstream end of life on April 1, 2027, and Redis 7.4 and later are no longer under the BSD licence. Valkey is BSD-licensed and maintained under the Linux Foundation. AccelerateWP 1.11-1 runs Object Cache on Valkey 9.1.
Only AccelerateWP’s own caches switch. The redis and alt-redis packages, yum install redis, any Redis you run yourself, LiteSpeed’s "Redis for users" and other applications on the server are untouched. Our December 2025 post on the move to Valkey planned this for Q1 2026 and described it as a switch of Redis itself. What ships is later and narrower. This post replaces it.
The update installs the alt-valkey package as a dependency. AccelerateWP then restarts each user’s cache once on Valkey, in the background and a few users at a time:
wp-config.php and every other website file stay unchanged. WordPress, the Redis Object Cache plugin and the PHP redis extension need nothing, because Valkey speaks the same protocol.All 25 server configurations in our automated tests passed. We don’t expect a visible change in site speed. After a mixed read-and-write load in our tests, the cache used 16% less memory than Redis 6.2.

redis-server process for each user becomes valkey-server. Its files in ~/.clwpos/ are now valkey.conf, valkey.log, valkey.pid and the socket valkey.sock, while redis.sock stays as a link to valkey.sock so sites keep connecting through it. Update your monitoring and scripts if they look for AccelerateWP’s redis-server, redis.pid or redis.log.To check a server, list the cache processes, then ping a user’s cache:
ps aux | grep -E 'valkey-server|redis-server'
/opt/alt/valkey/bin/valkey-cli -s /home/<user>/.clwpos/redis.sock ping
The second command answers PONG. Each cache switches at its own restart, so for a short time after the update some users are still listed with redis-server. If some stay there, run clwpos-daemon rolling-restart --replaced-binary-only as root. It resumes the switch and leaves the caches already on Valkey alone.
alt-redis stays installed, but AccelerateWP no longer requires it, so dnf autoremove or apt autoremove may offer to remove it. Keep it if anything else on your server uses it.
On other servers, the regular update installs everything.
alt-valkey: apt-get update && apt-get install accelerate-wp.alt-* exclude). They can’t install 1.11-1, and that setup isn’t supported. Remove the exclude before you update.There is no setting to stay on Redis. To go back, downgrade the package: yum downgrade accelerate-wp, or on CloudLinux for Ubuntu 22.04 apt-get install --allow-downgrades accelerate-wp=<version>. The caches then return to Redis.
On CloudLinux 7 to 10:
yum update accelerate-wp --enablerepo=cloudlinux-updates-testing
The minimum versions are accelerate-wp 1.11-1 and alt-valkey 9.1.3, which installs automatically.
CloudLinux for Ubuntu 22.04 has no beta channel. Ubuntu servers get 1.11-1 when it enters the gradual rollout. To take it at the start of the rollout, enable the rollout-bypass repositories:
sed 's/^# *//' /etc/apt/sources.list.d/cloudlinux-rollout-bypass.list.disabled > /etc/apt/sources.list.d/cloudlinux-rollout-bypass.list
apt-get update
apt-get install accelerate-wp
To leave the bypass, delete /etc/apt/sources.list.d/cloudlinux-rollout-bypass.list.
AccelerateWP 1.11-1 works on CloudLinux 7, 7 Hybrid, 8, 9 and 10 and on CloudLinux for Ubuntu 22.04, with cPanel, Plesk or DirectAdmin. Pricing doesn’t change. Object Cache stays free of charge with Single Hosting Account (Solo), Up to 5 Hosting Accounts (Admin) and Unlimited Hosting Accounts (Shared Pro) licences, as it has been since September 2024.
If you want Valkey for your own applications too, and your Redis is used as a cache with no data kept on disk, the switch takes three steps:
dnf install alt-valkey (yum on CloudLinux 7, apt-get on CloudLinux for Ubuntu 22.04).maxmemory) from redis.conf into /opt/alt/valkey/etc/valkey/valkey.conf.systemctl disable --now redis (redis-server on Ubuntu, alt-redis if you run that), then systemctl enable --now alt-valkey.Applications keep working if they connect to the same port or socket. This doesn’t apply to Redis 7.4 or later, Redis Stack or Redis modules. With data on disk the switch is one-way, because Redis can’t read data that Valkey has written.
If something on a server doesn’t switch as described, contact CloudLinux support.
Nikolay Galov
Comments