Metaframe XP WAN users kicked off 1

[mcputer]

I don’t think I am the first to have this problem, and I am sure I won’t be the last.

I have a combination of WAN and LAN Citrix and Microsoft TS clients. All of the WAN clients are Cirtix,

My network to the main office is 10Mb Wireless, rock solid. Outlying offices ADSL 2.5Mb mostly solid but with the usual ADSL problems.

My problem

The WAN users are kicked off the network apparently at random. There are no load or time constants that I can see. A site with 35 clients will work fine for a number of days and then groups of 3 or 4 will get the Citrix network difficulties 30 second timeout and auto reconnect message. All other users at this site continue to work just fine and dandy. Later another group of 4 or 5 will get the same message; and again everyone else woks just fine.

We are using Metaframe XP Windows 2000, 3 terminal servers with load balancing, Office 2K and a database front end to SQL, and of course the Internet.

LAN users with Citrix have no problems and there are no problems with the Microsoft client.

I have increased the Keyboard and mouse timeout to 300 from 0, enabled fragmented packets on the Sonicwall devices.

Any ideas?

Rob

 

[timeslice]

Hi,

I have the same problem here in Brazil. Our Citrix structure is the same (5 Windows 2000 Servers with load balancing, Metaframe XP, etc.). This problems begins to occurs after our Domain Migration - we had a Windows NT Domain, now it´s a Windows 2000 Domain.
Can you say if changing the mousetimer and keyboardtimer make any difference?

Marcelo

 

[gbiello]

Have a look at the bottom half of CTX708444 on Citrix's support site.

You may wish to try some ping tests, hitting the public IP of these remote sites both when things are fine, and when the problem is manifesting itself. See if there are any differences that could indicate a WAN issue.

-gbiello

 

[mcputer]

I keep a tight monitor on the WAN and it doesn't seam to coincide with any WAN problems.
The other thing is it doesn’t affect everyone at the same time. Out of 30 users it may only affect 2 –5.

Rob

 

[neilcresswell]

Have you tried increasing the TCP timeouts to allow degradation of throughput and latency...
Reg Changes need to be made to both the clients and the server for the changes to be effective. The problem may only be momentary which is why some of the clients fail, they may have established a session at that moment.

If nothing else, try it out on one machine that is experiencing the problems, what harm can it do..

(from CTX757449)
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
"TcpMaxDataRetransmissions"=dword:0000000a

(and from ctx708444)
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
"KeepAliveTime"=dword:0000ea60
"KeepAliveInterval"=dword:000003e8

 

[mcputer]

Thank you , all good stuff. I will be installing SP2 later this week and I will be sure to give the keepalive a boost.

 

[mcputer]

I have implemented the action in CTX575449 and things are a lot better.

Thanks to everyone for their help.

I will now apply SP2 for Metaframe XP and I think I will be home a dry.

 

[bran2235]

This may sound stupid but give it a shot! We had the same problem with connectivity in one of our remote offices... Our CTX servers are here in Richmond, VA and our users in Houston, TX would all of the sudden get kicked off/out... there connection would just cease!! We tried everything! Had about 20 people come in to look at inside wiring, infastructure, had our WAN provider remap PVCs, etc... We couldn't beleive it but the problem turned out to be a cheap 10/100 NetGear Hub that was in that remote office. We upgraded it to a Managed Cisco Switch and, like magic, the problem just disappeared!!! Give it a shot!!! We couldn't beleive it!!!
Good luck,
Brandon

 

[mcputer]

I have also gone down this road. We use 3com 24 port hubs and 3com 5 port switches. Good quality stuff, but they can fail.

I will keep this in mind. Thanks for the sugestion.

Rob