Return back to Proxy Cache main section
The Caches Center and the cache status page show "Unsupported storage type using multiple CPUs" for a disk cache that the proxy does not use because of its multi-CPU configuration.
¶ Why
To use several CPUs, the proxy runs several workers (processes). The cache types do not behave the same way:
- A rock cache is shared by all the workers.
- An aufs, ufs or diskd cache cannot be shared: it is used by one worker, the process chosen in the settings of the cache (process 1 by default).
The message appears when an aufs, ufs or diskd cache is assigned to a process that the proxy does not run, for example process 3 when the proxy runs 2 workers. The Caches Center shows the process of the cache and the number of workers next to the message, for example (Process 3 / 2). Such a cache is left out of the proxy configuration: nothing is stored in it.
This usually happens after reducing the number of CPUs used by the proxy, or when moving a configuration to a server with fewer CPUs.
¶ How to fix it
- Assign the cache to an existing process: in “
Caching” > “Caches Center”, click on the cache, choose a process that exists in the Process field, save, then click Apply configuration. - Or use a rock cache, shared by all the workers: create a new cache folder in the Caches Center (in multi-CPU mode, a new cache folder is a rock cache), or enable the Central Cache. Then remove the unsupported cache.
- Or reduce the proxy to a single CPU: with one worker, every cache type is used.
¶ Good to know
- Before October 2026, every aufs, ufs or diskd cache was reported as unsupported in multi-CPU mode and none of them was used. They are now used by the worker of their process.
- An aufs cache only serves the traffic of its own worker. With several workers, a rock cache gives a better hit ratio, because every worker shares the same objects.
- In
/etc/squid3/caches.conf, an unsupported cache appears as a comment:# cache "name" (directory) is assigned to worker 3 but Squid runs 2 workers: not used.