Here's a puzzle for you. I have found no helpful information on this anywhere.
We have 2 exchange 2007 CCR clusters and on both clusters the number of emails stuck in the send queue builds until I fail the clusters over. I am finding this is necessary about every two weeks now.
The exchange servers are patched to the latest SP1 and rollups. They are on windows server 2003 SP2 with the latest patches as well.
We use spotlight on Exchange which shows me the send queue on the information stores, but it agrees with the number of emails in performance monitor on each exchange server so I know this is accurate.
Noone is complaining that their emails are not being sent however, which I find odd, and when I fail the servers over this queue clears immediately.
Any ideas as to why this is happening and how to stop this behavior?