Excel Link after timing out throws error message when trying to log back in

[Ian Leslie]

Hi,

I'm having a problem where if the user leaves their Excel open for a time it will log them off from the Excel Add-In, when the user then tries to log back on via the Excel Add-In an error message is thrown, the user is accessing Controller and the Excel link through a RDP session, using Excel 2010

Error occured at 7/23/2014 11:48:12 AM in Microsoft Office 2010, Error No=5, Source=FrangoDirect.ConnectionDirect.GetUserInfo#ControllerProxyClient, Description=System.Web.Services.Protocols.SoapException: Server was unable to process request. ---> System.IndexOutOfRangeException: Index was outside the bounds of the array. at Cognos.Controller.Proxy.ControllerReportNetService.RepNet.GetUserInfo(String sGuid, String sUser, String passportId) at Cognos.Controller.Proxy.CCRWS.GetUserInfo(String sGuid, String sUser, String passportId) --- End of inner exception stack trace ---

[IBMrc]

Errors when users leave their sessions idle make me immediately think of timeouts. The main two timeout settings are:

• Cognos Configuration
  
• IIS

I suspect that it's most likely being triggered by the IIS timeout (similar to this known issue: http://www.ibm.com/support/docview.wss?uid=swg21605585).

=> The solution is probably to increase the IIS timeout (I suggest to 120 minutes) by following these instructions: http://www.ibm.com/support/docview.wss?uid=swg21365061

Good luck! :)

[Ian Leslie]

Thanks, currently the time-out is set to 60, I'll try the 120 setting as you suggest and let you know if it's successful.

[Ian Leslie]

Hi IBMrc,

The changing of the App pools didn't make any difference, however I've been able to solve half the issue.

If the client starts the rdp session with the colour set to 32 bit, the domain log on when F9 is pressed after timing out will not reveal itself, after a wait the error occurs, when the rdp session is started with 16 bit colour then hey presto the domain pop up appears after pressing F9, the client is able to log back in, this is a good workaround as this means the client doesn't have to go through the effort of closing and reopening from scratch.

However, if the client times out and they want to log back in via the Controller Excel Add In menu the Log In menu item is available, the db selection appears, but after selecting the db it throws the error, this is through the rdp session.
If I try to replicate the issue by being directly on the server I can't as although it times out, the Log In menu item is always greyed out, so I have to log out, which asks me for the domain log in, which then logs me out, which in turn makes the Log In menu item available, the log off then being greyed out, I can then log in successfully

Any thoughts before I raise a pmr with IBM would be appreciated.  :)

My thoughts are along the lines of the rdp settings, so next step is to check what they are!

[IBMrc]

The description is a little hard to follow ;) but it's obviously something to do with user re-authentication (after timeouts).

Often customers are happy with workarounds, so if you only get the problem after (say) 60 minutes, then try increasing the timeout.

• You've already tried increasing the IIS timeout (to 120 minutes I believe). You can try increasing this to (say) 600 minutes (10 hours ... therefore they would not get a problem all day). Instructions are here: http://www-01.ibm.com/support/docview.wss?uid=swg21365061

• You also will need to increase the Cognos timeout. By default this is one hour too (3600 seconds) so try increasing it to 36000 (10 hours). I suspect that everything will be OK after doing both of these. Instructions are here: http://www-01.ibm.com/support/docview.wss?uid=swg21339043

In an ideal world, of course, it would be great to have the exact cause/fix (not just workaround), but (being realistic) finding the cause/fix would involve quite a lot of time/effort on the customer's side, and they may prefer to stick with an easy workaround.

p.s. very strange to hear how the RDP colour settings affected things. Are you sure that this was not just a coincidence? 100% sure you were testing 'like for like'?