I have 2 Exchange 2007 virtual machines. My topology is 1 Exchange server running cas/mbx roles and another exchange server running mbx role. The VM's have 4 virtual processors and 6GB of memory and do not show signs of capping resources. I am getting reports of users being disconnected from Exchange and making their mailbox unusable without restarting outlook. I am seeing event 2262 "deadlocking" in the event viewer and wondering if this could be the reason for the disconnects. The users are using http over rpc in a mixed cached/non cached mode. My goal is to no longer receive the deadlocking event as well as resolve the disconnect issue.

Service Pack and Rollup level? Have you run ExBpa against the servers? 6GB seems to a little tight. Multi-Role Servers should have a min. of 8GB. Have you sized these servers correctly? Personally, for a multi-role server, I would have 16GB- regardless of how many users you have unless its "very" few. http://msexchangeteam.com/archive/2006/09/25/428994.aspx http://technet.microsoft.com/en-us/library/bb738124(EXCHG.80).aspx

I've ran ExBPA and it didn't report any memory issues. I usually have some memory free, not much, but some. I will increase the memory from 6 GB to 8 GB as you suggested, but let us assume that will not fix the deadlocks due to the fact that I'm barely swapping if at all due to my memory usually haveing a couple hundred KB free. What can I do to properly diagnose this issue and figure out exactly what is causing the 2262 event ID?

Well, diagnosing is a little difficult. I think you would need to engage Microsoft or an IIS expert. Ensure you have the latest SP and RU applied as there have been a number of Exchange fixes for these sort of issues.