On 04/06/12 03:40, Stuart Bishop wrote:
On Fri, Apr 6, 2012 at 3:51 AM, Gary Poster<gary.poster@xxxxxxxxxxxxx>
wrote:
I'm seeing these pretty consistently on parallel test runs. It seems to
affect both test environments. I filed bug 974632.
Have we considered dropping Memcached? Last I heard, it was only being
used on the front page for cached statistics. For this sort of use, we
could easily replace the IMemcache implementation with something with
less moving parts (simple table in PostgreSQL, global dictionary per
appserver, no storage at all - do we actually gain much with the
cache?). The interface was deliberately simple to allow us to do just
this if Memcached didn't suit our needs.
It would be fine by me, but since Brad and Graham seem to have found the
solution to this particular problem, dropping memcache is out of scope for
my squad right now.