Since it seems to happen sporadically and only on our busy production server, I wonder if it isn't memcached starvation.
We should devise a test for that condition, perhaps using a very small memcached setting.
Since it seems to happen sporadically and only on our busy production server, I wonder if it isn't memcached starvation.
We should devise a test for that condition, perhaps using a very small memcached setting.