Good afternoon, all!
Just finished updating my Exchange 2013 servers and ran across a circumstance I haven't seen before.
Topology: two datacenters with a Mailbox/CAS server in each, both are members of the DAG. These connect to two Edge Transport servers, one in each datacenter. The Edge - Mailbox/CAS server connections are cross-connected so any Edge can connect to any Mailbox/CAS server.
Ran updates on the two Edge servers - no problem. Used this document as a guide, starting at "Performing Maintenance on DAG members: for the two Mailbox/CAS servers. Started with the remote server, followed the guide, tested email during the procedures, mail flowed normally.
When I updated the primary server at our main datacenter, outbound mail continued to flow normally. However, inbound mail was held for about 20 minutes while the primary server completed Windows Updates and rebooted. As soon as I took the server out of maintenance mode, the inbound mail was delivered to the mailbox. I did check and the database was hosted and mounted on the secondary server. I don't have any custom receive connectors that may be
There were some inconsistencies in the output of Get-ServerComponentState for each server. However, when I tried to set the components active, they immediately went to Inactive. The primary server components all restarted after the patch reboot, and the cluster, after yelling at the reboot, shows both servers as up.
I can handle a planned outage of 20-30 minutes or so if necessary. However, the whole point to creating a DAG and having the redundant servers is precisely so that a mail outage won't happen. If this patch session is a gauge, it seems that inbound mail from the Internet may be delayed or bounced if an outage occurs.
What might I be missing in the inbound mail path? Are there other components that I should be checking for proper mail failover?
Thanks for looking!
Gregg