I've experienced this a couple of times now. I've followed this guide at SP2-upgrade, and at that point it seemed that during step 5.. the passive node suddenly takes ownership and starts being a bit online, information store was in failed state, and hence my mailbox databases was offline as well - Very annoying. And preferred owners are set to both NodeA and NodeB - Any hints of what I'm doing wrong here?
Today, when upgrading to SP3, it did exactly the same thing - And when I came to do step 8, it complained - So I figured, hey... must be the same issue as last time. I tried using the EMC to move the mailbox server role back to NodeA - But it complained about version differences (naturally). So instead I used Failover Cluster Maintenance to move all cluster applications to NodeA - That worked, and it went online smoothly. I was now able to continue from step 8.
Esther Fan, MSFT: Thank you for your comments. For these kinds of questions, please try the following forums:
TechNet: http://social.technet.microsoft.com/Forums/en-US/categories