Exchange 2010 Service Unavailable during SMTP transactions. Exchange Server is not accepting any incoming emails

I must say that my dear old friend Murphy paid me a visit this week. For those of you that are unaware of Murphy, you can get more information here . Basically Murphy's Law says that "Anything that can go wrong will go wrong"

Anyway, not to get side tracked. This week, one of my legacy Exchange Servers (Exchange 2003) broke down. Or that was what it seemed superficially. I am still trying to work through the migration of my Exchange 2003 servers to Exchange 2010. I still have a handful of users left on my legacy server and so am trying my best to move them across. As it happened, on this day, users on my Exchange 2010 started receiving Delay notifications when sending emails to users on Exchange 2003. Users on Exchange 2003 were also receiving the same when they tried sending emails to anyone on Exchange 2010 or to those outside the organisation.

I checked the queues and found that the Exchange 2003 queues were increasing with time and Exchange 2010 queues with emails destined to Exchange 2003 were suffering from the same fate.

My first reaction was that network issues might have cropped up, causing havoc between Exchange 2010 and Exchange 2003. Since Exchange 2003 was configured to deliver emails using Exchange 2010 for out-of-organization recipients, this theory  made perfect sense. But then when I checked the network connectivity between the two servers, there did not seem to be any problems :(

I then looked at the SMTP logs on Exchange 2003. To my surprise I found that one of the things logged was "Service Unavailable".  This meant that the Exchange 2010 CAS server was not receiving emails. I opened up a command line and tried connecting to the Exchange 2010 SMTP service, but after the handshake, I also received "Service Unavailable".

I RDP'd to my Exchange 2010 CAS server and checked the event logs. To my surprise I found that it was throttling the incoming email transactions. This was because the used diskspace on the disk that stores the queues had gone beyond 94%. This is known as BackPressure.

The simple solution was to find and remove all unnecessary files on the Exchange 2010 CAS server to bring the free diskspace up. As I looked through the eventlogs I saw Exchange reporting that the load on the resources had lessened and as such the limitations on the incoming queues was being removed.

Phew!. I forced the messages in the retry queues in both Exchange 2010 and Exchange 2003 and watched as the numbers reduced.

For those of you that would like to read more on Exchange 2010 Back Pressure, you can click through to the following articles
Technet - Understanding Back Pressure
MSExchange.org Back Pressure in Exchange 2010

I would recommend the MSExchange Article ;)