Optimising Win2K Terminal Server 1

[glynnj]

What can I do to optimise a Win2k Terminal server.

The Server -
Dell Poweredge 1550
Dual Processor 1266mhz
2GB RAM
Windows 2000 Server SP3 (Latest patches installed)
IE5.5 SP2
100MB Net Card

We have 20 users (Win2k Pro) connecting to the server via 100MB network. Periodically during the day several users have connection problems. For example, their terminal services display will freeze or react very slowly. This happens at different times during the day to different users. I have swapped users machines and network cables. There does not seem to any serious amount of network traffic. If the user reconnects the problem goes away but will come back again. Has anyone had a similiar problem or know of a fix?

Thanks and regards,

Joe

 

[electraboy]

Try Symantec FixIt 4.0 It is impossible to make anything foolproof because fools are so ingenious.

Murphy's Laws​

 

[electraboy]

Sorry,
The vendor is not Symantec, but Ontrack It is impossible to make anything foolproof because fools are so ingenious.

Murphy's Laws​

 

[millap]

I'm having the same problem with two 2000 boxes but in a slightly different setup. We have 3 Citrix Metaframe boxes (2 w2k, 1 nt4 TS). We run remote desktops on the w2k boxes and from time to time during the day users get 'stuck' in whatever their doing (say, opening new windows, logging on, or simply clicking on START). One w2k box is a Proliant 6000 Quad PII 450 with 2mb L2 Cache and 1.5gig Ram and a 100mb NIC the other is a Proliant DL360 Dual PIII 1.0ghz with 1gig Ram and a 100mb NIC. Both of the servers are patched up to W2k SP3 and Metaframe FR1 (we were suffering the same problem before the SP3 install).

On times of slowness, Task Manager show's cpu utilisation as minimal, paging from memory is low/non-existant and disk access is minimal. Leaving it for long enough (sometimes 15 minutes) clears the problem. Other times we log users off. This clears the problem and allows users to log back on again. Network traffic isn't overloading these servers, I run MRTG to monitor the info going back and forth. It shows no more than 20k per second WELL within the parameters of the cards.

The DL360 can handle 30 users and only ever seems to have a maximum of 20. The 6000 can handle 60 users and only ever seems to have a maximum of 30.

I've been frantically looking for a solution to this both on the MS and Citrix KB but can't seem to find anything.

Sorry your suffering the same problem Joe, but I'm glad in a way it's not just our setup. I'd love to hear about any performance enhancements too

Andy

 

[CitrixEngineer]

This looks like an issue I saw ages ago with mapped network drives. To fix this, I followed the great advice in this article

http://www.thethin.net/maxmpxct.cfm

Hope this does the trick CitrixEngineer@yahoo.co.uk

 

[JWayne]

I have had this similar problem as well and what I have noticed is a degradation in the internet connection itself. I experience dropped packets, increased ping response times etc. Unfortunately, this is the internet unless you have leased lines. Keep a log of ping results from your router locally to the remote router. You may find that when your users are experiencing problems is when the internet is suffering. You can always send these ping results and trace routes to your ISP and they can see where the bottleneck is coming from.


Good luck

 

[glynnj]

At least I'm not the only one with this problem, Andy. I'm going to try the fix suggested by CitrixEngineer. I am going to set the following on the Terminal Server.

HKLM\System\CurrentControlSet\Services\LanManServer\Parameters
MaxMPXct - 1024
MaxWorkItems - 4096
MaxCmds - 125

If all goes well, I will make the same changes on the DC and mail servers. I am hoping to get this done over the next few days, but given that this is an intermittant problem, it will probably be a few days before I will know if the problems has been fixed.
Check MS articles Q271148, Q191370, Q232476.

What have you tried so far Andy?

Regards,

Joe

 

[millap]

Hi Joe,

I'm going to try the same as you on the one NT4 TS we have to see if that's the root of the problem. The other Terminal Servers are 2000 and Q232476 specifically says it's not for 2000 boxes, whilst Q271148 specifies to install SP2 for 2000 (which is already done). Aside from that, I've not been able to try anything else, as we've both discovered, the boxes aren't under any undue strain so throwing more hardware at it wouldn't solve the problem. As it stands this is the closest I can see to a solution.

Good luck with your end
Andy

jwayne: We have 512k PVC's running into our main depot (2mb PVC) with no more than 15 users on each. Traffic over these links never reaches over 10k per second so I'm pretty sure the problem isn't there.

 

[glynnj]

Andy,

This problem doesn't really seem to be a TS issue alone. Q191370 suggests making these changes on all servers accessed from the TS server, so if you have a file server, you might need to apply these changes to it as well. Also these changes do apply to Win2000 server, if you look at q271148. I know MS articles often contradict each other but I assume the latest article is correct.
Let me know how you get on.

Good luck,

Joe
PS SP3 is available for Win2000 server.

 

[millap]

Joe, all of my 2000 servers are already SP3'd which kinda tells me the problem should already be fixed as per the Q article (!).....

 

[CitrixEngineer]

glynnj - I hope you read the second article on the link I provided - or your Windows 9.x clients (if you have any) will have severe problems!

5000 might be a safer setting for MaxWorkItems than 4096 (and you would, of course, also have to change MaxMPXct to 1250). I found that it wasn't the terminal servers that required this setting as much as the servers that held Home/Profile directories and file/printer shares.

Hope this helps CitrixEngineer@yahoo.co.uk

 

[glynnj]

Thanks CitrixEnginer. I had read the article but we only have Win 2000 clients so we should be ok. I have changed the MaxWorkItems and MaxMpxCt on all servers accessed by terminal services, including the Mail server and file server. No problems so far but it will probably be a few days before I can be sure the problem is fixed.

Regards,
Joe

 

[glynnj]

The changes to the registry seem to have fixed my freezing problem. I used the following settings -

LM\System\CurrentControlSet\Services\LanmanWorkStation\ParametersMaxCmds (DWORD) 1024 (Decimal)

LM\System\CurrentControlSet\Services\LanmanServer\ParametersMaxWorkItems (DWORD) 20480 (Decimal)
MaxMpxCt (DWORD) 5120 (Decimal)

3 Windows 2000 Servers (SP3)
Terminal Services
File Server
Exchange Server

The changes need to be made on all servers (and rebooted). All clients are Win 2000 professional.

Also found another cause of problems on TS. If Office XP is installed with alternate user input, it causes Outlook to stop responding. MS Articles Q300904