Exchange 2007/2010 Issue with IMAP / TMG Redirection Rule
Posted on 2011-09-30
We are having issues with IMAP clients (various clients on various OS’s) connecting to Exchange 2007 CAS and mailbox servers when we change our TMG rule to redirect clients to our 2010 CAS in preparation for an eventual migration to Exchange 2010. Currently there is a single external entry point for all IMAP clients over port 993 pointing to our TMG server. Our Exchange system consists of TMG servers in an Array (two servers with a DB server) There are two Exchange 2007 CAS running an NLB and a single 2007 CCR mailbox server. Exchange 2010 has two CAS running an NLB with a 2 node DAG. When we switch the TMG rule to point to the Exchange 2010 CAS NLB we get a variety of intermittent client RECEIVE (fetch) connectivity issues ranging from poor performance and IMAP synchronization to on some clients, inability to connect. Please note all mailboxes are on Exchange 2007 we are just changing the TMG rule for IMAP clients to point to the 2010 NLB. Testing was also done redirecting to a single 2010 CAS with similar results.
• Thunderbird reported this Alert
“An error occurred during a connection to <OurDomainName>:993. SSL received a record that exceeded the maximum permissible length. (Error code: ssl_error_rx_record_too_long)”
• Reinstalling or upgrading the IMAP client seemed to have helped in some instances with this and some other performance / connectivity issues.
• Outlook, configured as IMAP; reported synchronization issues.
• In other tests, we've had IMAP clients reject the mailbox password and continuously prompt for a password without locking the users AD account due to bad passwords.
I opened a case with MS support and they analyzed the IMAP logging files, along with everything else on TMG, DC’s and Exchange servers without finding any issues with these servers. The IMAP clients are configured correctly as they can connect once the rule is reverted back to point to the 2007 NLB.
One would assume that changing a TMG rule to point from an existing functioning exchange 2007 server to an Exchange 2010 server would not generate widespread intermittent client connectivity issues, but it seems to in our exchange configuration. Please note that we are NOT moving the mailboxes just a TMG rule.
Has anyone come across or experienced this? Any known solution? Any and all information is greatly appreciated