I have a web cluster which is using RAM disks on both cluster nodes for caching.

The challenge is for users to be able to trigger a cache clear which will clear the contents from both of the RAM disks.

What I tried:
Using a GlusterFS, the RAM disks are both mounted to a common folder.

This works for caching but the clearing function (triggered by a Wordpress plugin) only affects the cache the post command hits.

The load balancer round robins connections so one is always missed.

Even though one of the caches clears, the cache remains on the other. Despite sharing a common folder the nodes are maintaining their own cache contents which are invisible to each other.

Is it just a matter of setting the same numeric based owner and permissions on both nodes for the shared cache mount point?

Otherwise, I'm definitely open to suggestions.

I would suggest you take a look at glusterfs per volume io cache config and implement "writebehind.flush-behind()"