Comment 4 for bug 1578401

This behavior is actually expected based on our current implementation. keystonemiddleware is actually using the memcached library directly and setting timeouts. keystone uses dogpile and sets a dogpile expiration and not a memcached expiration. This means that the data will expire like expected, but it won't be expired as far as memcached is concerned.

Other than appearing to be fuller than it actually is, I don't think there is too much impact. Even when a memcached timeout exists on an entry is will not actually be deleted from memory by memcached until it is accessed again or memcached needs the memory.

I've created a POC change in olso.cache that "fixes" this problem.