Dears,

We have outlook clients get disconnected randomly in branch sites from exchange 2013 and connect randomly back again...I suspect MTU value is the issue but how can I verify this? note that in main site (in which exchange is installed) outlook is working fine.

I did the below to troubleshoot this issue:

1. Pinging mail.domain.com works fine with some network delay
2. Trace route from client to mail.domain.com and it shows direct connection while it should not, they are in two different vlans (main site and branch site)
3. Test email auto configuration and it works even outlook is disconnected (which do not make sense)
4. Test accessing OWA, it worked but it is slow
5. Test creating new outlook profile, it worked but hanging .. seems got into timeout while configuring in the wizard

rom the behavior above it seems to me there is some network issues preventing stable connection for your outlook clients in branches to exchange and I am not sure if MTU is the issue.

• Edited by Ahmad K. Jayyusi Thursday, May 21, 2015 6:39 AM

Hi,

According to your description, I understand Outlook client randomly lose connection in branch site, however works fine in main site which Exchange server installed.
If I misunderstand you concern, please do not hesitate to let me know.

The causes for this issue have many things, therefore we need to narrow down more possibility issue.
Please refer to Sivakumars answer in below link: https://social.technet.microsoft.com/Forums/office/en-US/6d6a7959-3c8f-45f7-b5c3-af9769f5f322/client-outlook-disconnection-of-randomly-outlook-clients?forum=exchangesvrgeneral

1. Try and engage your network team and make sure they use good tools for network packet analyzers to analyze where are the packet drops happening. Of course you cannot get this tool setup when things are broken, so its your call to run this tool and leave it running when you think the problem COULD occur
2. When the issue happens - dump your RPC Client Access logs to another server and make sure to have a detailed log analysis done
3. Get a list of number active CAS RPC connections per CAS server. This will tell you if there any patterns specific to a CAS server -  http://www.mikepfeiffer.net/2011/04/determine-the-number-of-active-users-on-exchange-2010-client-access-servers-with-powershell/
4. Do a netstat command on the CAS Servers to see if the CAS server lost communications to the DC/GC's
5. When the problem happens, make sure to get the Domain Admins to run a DCDIAG and check the log in depth

Meanwhile, we can try to change the WAN MTU for troubleshooting. The default MTU sizes for different network topologies , for your reference:
https://support.microsoft.com/en-us/kb/314496

Thanks