If I could upgrade the ram in these systems, I would. They are maxed out.
I haven't heard back from the hosting company about how much money it would cost to upgrade to completely new servers.
I'll note that if anything all the advice I read on the web about sql cache was to *drop* its value - not raise it - to help with this problem. The problem apparently isn't RAM here. It's that two SQL threads are in contention and are crashing into each other. If anything the advice said a larger cache created a larger opportunity for that crashing to occur and that reducing the RAM dedicated to the sql cache minimized the chances of this happening (because the cache would clear before the collision occurred). But they said in general this event was indicative of something wrong with the code, that there was that potential for two threads to hit the same area at the same time. That the code should ensure one process finished before the next began.
So in this specific case I don't think more RAM will help, since everyone is saying to reduce the RAM used, not increase it.
See here and then follow the link to the bug report -
http://serverfault.com/questions/176465/mysql-thread-get-stuck-in-end-stateAgain removing the cache made the problem go away. So putting all load on active memory fixed the problem. So to me that doesn't sound like RAM was the bottleneck.
