Major Description
We had added a feature to spymemcached based on a Couchbase user's request to detect hung processes. This tries to complete a noop before doing auth.
It's fine against memcached/ep-engine in all cases, and it appears to be fine for ascii (where there is no authentication and it fails back to the version command), but moxi does not seem to allow auth after the noop. This may be because it's expecting the first command to wire it up to a downstream for the "gateway" moxi?
We're going to search for a workaround, but I wanted to make sure this issue was known.
See also:
https://code.google.com/p/spymemcached/issues/detail?id=272&thanks=272&ts=1364702110
and
https://github.com/mumoshu/play2-memcached/issues/17