With all of the timeout settings on the WinServer 2012 R2 box (Group and User) set to "Enabled|Never", or alternately to "Disabled", the MRD session on my macOS Yosemite/El Capitan terminates at ~60 seconds.  On a stopwatch in various configurations I see timeouts between 54 and 72 seconds, but that is almost certain slop somewhere in the interaction between host and client.

Host: WinServer 2012 R2 English

Client: macOSX v.10.10.4, but this has been an issue since at least v.10.8 and the MRD which was current in late 2014, and on many other macOS machines which I interact with.

This happens whether the client machine is inactive on the Remote Host with the MRD window in the foreground with active processes running, or with the MRD window minimized or in the background with idle or active processes.

[EDIT] No timeout problems on this host when using MRD on Win7 or Win8, or in iOS on an iPad or iPhone.  So I do not believe that it is a problem on the host, rather it is a settings problem on the client - but there is no place to set an idle timeout in Yosemite!

TCPIP idle timeouts have also been tweaked to no effect, and now we have customers reporting the same issues running RD Client in Win8 x64 and Win7.  Still no problems on iOS8 or iOS9 versions of RD Client.

Hi,

I forwarded this issue to product team. In our test the issue occurs 2 times in 20 connections. How frequently the issue occurs on your side?

Meanwhile please understand that we may need to collect more information for further analysis. I'll update my reply when I get response from product team. Thank you for your time!