?
Solved

Terminal Services Slowness

Posted on 2006-05-28
37
Medium Priority
?
1,256 Views
Last Modified: 2013-11-21
We have a site running two brand new HP ML350 g4p servers. A file server with 2.5 GB of RAM and a terminal Server with 3.0GB RAM. Both with Xeon 3.2 Ghz processors in them. The site is an engineering company that runs many software applications, with differing levels of external support, concurrently. We were employed to take over the site when the previous terminal (2 x Xeon 1Ghz with 1GB RAM) and file servers (1 x P4 2.4Ghz with 1GB RAM) started to have major hardware failures after 3.5 years of use, they were generic clone boxes. Post implementation, the staff are saying that the new system is slower than the old system. In response to this, we have individually re-build the profiles that the 20 thin client users log on to. We have tried having the profiles on either the terminal or file server. No speed difference.... we have re-build both machines, no speed difference, we have tested the cables, no speed difference. We have put in a Linksys Gigabit switch, no speed difference, we have tried an additional processor in each machine and more RAM, page file use was down under full load, but no speed increase! Can anybody suggest a new approach so we can attempt to get a solution to this problem???
0
Comment
Question by:comptech_engineering
  • 12
  • 10
  • 5
  • +4
35 Comments
 
LVL 78

Expert Comment

by:Rob Williams
ID: 16781069
Have you looked at anything as simple as name resolution ? I have seen DNS issues slow down machines. This will be most noticeable at logon, but can extend to working with any file shares, or remotely stored profiles. If in doubt try running Windows netdiag.exe and see if there are any issues that show up.
0
 
LVL 1

Author Comment

by:comptech_engineering
ID: 16781176
We have checked DNS. No errors in the error log. We will try this file though. Is it part of Windows or do we have to download it from somewhere?
0
 
LVL 78

Expert Comment

by:Rob Williams
ID: 16781248
It is part of the Windows Resource Kit or you can download just that file from:
http://www3.ns.sympatico.ca/malagash/Downloads/Net/netdiag.exe
Try running on one of the terminal servers. It is a DOS based utility, though it will create a test file output in directory in which you place it.
0
Get free NFR key for Veeam Availability Suite 9.5

Veeam is happy to provide a free NFR license (1 year, 2 sockets) to all certified IT Pros. The license allows for the non-production use of Veeam Availability Suite v9.5 in your home lab, without any feature limitations. It works for both VMware and Hyper-V environments

 
LVL 1

Author Comment

by:comptech_engineering
ID: 16782053
We ran your diagnostic as suggested on both file and terminal servers. The terminal server passed perfectly without any dramas. The file server reported the following error...

"the procedure entry point netinfo_isforupdate could not be located in the dynamic link library. DNSapi.dll"

Does this suggest anything to you? Have you seen this before?
0
 
LVL 10

Expert Comment

by:bbrunning
ID: 16782927
What OS are the client computers running?
0
 
LVL 3

Expert Comment

by:ykamp
ID: 16783125
I agree about DNS . If you don't have the proper setup you will probably get delays. Have you set up WINS server? I've seen networks working really faster after a wins server is installed and properly configured on the primary domain controller.

I also suggest to rebuilt the users . Once after a failure on my RAID controller I had to set up my domain controller from the beginning. After the setup I thought that it was better to recreate the exact same users from the documentation I had and then without any changes to the clients let the network work as it was before as the were no changes. We used the same domain controller with the same domain , the same server name , the same ip address etc. I thought that when the clients try to logon they wouldn't find any differences so logically there wouldn't be any problem. When we tried we find out that there were delays when the users tried to logon and when they tried to start any server based applications. After trying many changes, with no results, to the domain controller we tried to add the clients from the beginning to the domain and create new users. That solved everything. I think that in the event viewer of the domain controller there were no error messages at all but I am not sure right now.

I am not sure that this help but I suggest you try it on test pc or on your pc. Remember disconnect the pc from the domain. Delete all domain user files that are stored locally (remember to take a back up) and let the pc works as a stand alone pc. Then create a new user and reconnect the pc to the domain from the beginning.

This happened about two years ago so I can't remember the exact reasons that caused the problem but I think it was an authorization issue.

Good luck
0
 
LVL 78

Expert Comment

by:Rob Williams
ID: 16783873
The "the procedure entry point... DNSapi.dll" is usually caused by using the wrong/older version of netdiag. Which version did you use? The reason I have the one on that FTP site is to have the current one available. I have had problems with some versions. I just tried the one above on a 2003 TS and a 2003 SBS and it worked fine. If you are running a different O/S you might want to check other resource kit versions. By the way you don't need to use the "setup" version. Simply copy the netdiag to a temp folder and run from that folder.
0
 
LVL 1

Author Comment

by:comptech_engineering
ID: 16786111
The computers are Compaq & HP thin clients, therefore are running a mixture of Windows CE and Linux.
The DC is a standard SBS install, so WINS is setup and as far as i can tell is working.
We have reset the thin clients back to factory standards, and rebuilt all the profiles. The next thing we planned to do was to setup terminal services roaming profiles and set gp to delete cached profiles. I don't know if this will fix things but i've read this good practice for a terminal services environment.

We simply downloaded netdiag from suggested location, copied exe to desktop and doubled clicked it. We are running the latest updates for MS SBS 2003 so we are on the latest version.
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16786877
1. Turn off anti virus.  Especially Symantec.  I use Trend.
2.  What printer drivers are you using.  The Laser jet 4 driver is one of the most stable.
3.  What applications are these people using, and if you installed microsoft office did you install the special TS add on for it?
4.  http://www.brianmadden.com/content/content.asp?ID=15
5.  http://www.brianmadden.com/subject.asp?ID=12

6. What do these people mean by "slow".  Do it mean 1 one-half second pause every time they click the start button before the menu appears, or a fraction of a second delay in screen updates during a menu scroll, or is it more like a 15 second delay when printing, or a unusually long log in process, or clutiness with the screen - they scroll and it freezes for two seconds and then catches up.

thanks,

-gsgi
0
 
LVL 78

Expert Comment

by:Rob Williams
ID: 16790606
>>"We simply downloaded netdiag from suggested location......."
I have had problems with version mis-matches as mentioned, you might need to do some digging to find the right one. I checked the MS site and the newest one doesn't include netdiag. However, if everything is OK on the terminal server, chances are your network and DNS have been configured similarly on different systems, so it is not likely the source of your problem.
0
 
LVL 1

Author Comment

by:comptech_engineering
ID: 16829786
How do we establish the appropriate version, via trial and error? and what other location would we best use to find the appropriate version? Currently the system is on a domain but doesn't use roaming profiles? would this help at all? or detract from the speed in your experiences... we are getting desperate if anyone out there has any more suggestions...
0
 
LVL 78

Expert Comment

by:Rob Williams
ID: 16829947
>>"How do we establish the appropriate version, via trial and error?"
Sad to say that's how I was doing it. I found I was getting .dll errors on numerous O/S's so when I found the one I referenced earlier I choose to upload and keep it handy. Seems to work on all versions I have tried. However, since your other system is configured properly I am guessing you would be consistent and that is probably not the problem.

Roaming profiles I don't think would improve performance at all and usually slow down the logons.

I don't have any other ideas, except possibly some monitoring to see if possibly a virus or spyware is "chewing up" CPU cycles or network bandwidth.
You might want to try putting a '0' point, pointer question to this one in the Windows Server 2003 forum, to see if you can stir up any new ideas.
http://www.experts-exchange.com/Operating_Systems/Windows_Server_2003/
Pointer questions:
http://www.experts-exchange.com/help.jsp#hi262

0
 

Expert Comment

by:vally7
ID: 16858870
Hi,

I'm posting this here as we appear to have the same issue with an HP ML350 G4p running SBS2003 (with all the latest MS updates).  We've also upgraded all of the firmware and drivers.  Not only does Network access appear to be slow, but disk performance is erratic as well.  We have configured RAID5 on a smart array 641 controller.  If we copy a large file from the c to d drive, it is slow!

If the author of this post has any breakthroughs, please can you update this call.  I'll let you know as well.

Thanks
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16858966
What do you guys mean by "slow"?  Does it mean 1 one-half second pause every time they click the start button before the menu appears, or a fraction of a second delay in screen updates during a menu scroll, or is it more like a 15 second delay when printing, or a unusually long log in process, or clutiness with the screen - they scroll and it freezes for two seconds and then catches up.

Is it longer than expect log ins, and log offs?  Or are the specific apps slow to process?

vally7@: How long does it take to copy that file?  How big is the file?  What is the antivirus being used?

Turn off Hyperthreading and test.

-gsgi

0
 

Expert Comment

by:vally7
ID: 16861369
Hi gsgi,

OK, I can't give you any exact measurements at the moment but I, much like yourself I imagine, have built and worked with a fair few servers to know when things aren't right.

The thing to remember here is that this behaviour is erratic, last night for example, I vpn'd into the network and got a RDP connection to the server console to troubleshoot this problem.  There was nobody else logged into the server, I'd put the backup job on hold, so other than the standard SBS processes that run, the only other processor/memory hungry app that was running was Exchange.  I had my eye on the Exchange queues as well to make sure that this behaviour was not caused by large mails coming in.  

Sometimes I would press the Start button, and the menu would appear instantly, othertimes there would be a 5-10 second pause before it appeared.  Another example, I tried starting IE and it opened up immediately, I tried opening another instance of IE about 10 seconds afterwards and it took 10 seconds for IE to open.  I then did the same test 10 mins later, and both copies of IE opened instantly.  I'm using IE as an example, I also tried with more memory hungary apps, same results

The file size I copied from C to D was 345MB and it took over 5 mins to copy.  I also did the same test on the old lesser spec'd server, single P4 (can't remember speed), RAID 5, and 1GB RAM and the file took 20-25 seconds.   I did the same copy on my workstation (P4 1.80mhz 128MB RAM) from C to D and the copy took 25 seconds.

From the users point of view, I went to a workstation, logged in as a user, went to their mail (Outlook client), created a new mail and then tried to attach 2 .jpg files each of 4 MB.  It took between a minute and 2 mins for the mail message I'd just attached the .jpg's to, to become active and show the two attachments.  The same test on the old server and it takes 10-15 secs

Also when the save files up to the server, this takes longer than expected.  A small word doc (under 100kb) can take between 5 secs (probably normal) to 30+ secs to save to the server.

Things I have done to try and fix:-
*Ensured I have installed all of the latest firmware from the HP Support CD (downloaded latest version last night), e.g Smart Array Controller, NIC etc etc

*Ensured I have installed the latest Drivers from the HP Drivers CD (downloaded latest version last night).

*I have tried putting pagefile.sys on just the C drive, also just on the D drive, and also split it over the two drives.  Incidently I have tried altering the size of this as per Microsoft's recommendations.  I have tried resizing it to 1.5 and 2 times the amount of physical memory.

*We have run the Exchange Best practises tool and configured exchange using it's recomendations (we applied SP2 to Exchange when we first installed SP2).

*I'm with you on the turning off Hyperthreading, it's one of the first things we did.  I've had these type of problems with this in the past with Microsoft and Netware servers and they were caused by Hyperthreading being turned on.


There are only about 20 users on the network, the network speed is 100MBs.  We did not experience these probelms with the old server, which was of a vastly lesser spec.  The new server is twin Xeon, 2GB RAM and about 450GB RAID5 and like I say, it is fully patched in terms of Microsoft Patches, and HP firmware and driver patches.  

We've also been running performance monitoring jobs, and are not noticing anything untowards in terms of disk access speeds, virtual memory overloads, physical memory problems etc etc.  I have also been monitoring packets going to and from the server, and nothing untoward is happening there.  There's nothing wrong in the event logs either, other than a sharepoint error, but we know what is causing that and can't see how that would relate to the erratic server performance.

My colleague has spoken to HP and been through this with them, the only thing that we do not have up to date is the BIOS.  We are going to try updating that this evening after hours.

phew!  if you can think of anything I've missed or can try gsgi then let me know, your help is very much appreciated!!

cheers
0
 

Expert Comment

by:vally7
ID: 16861395
I meant to write:-

*We have run the Exchange Best practises tool and configured exchange using it's recomendations (we applied SP2 to Exchange when we first installed SBS 2003) :-)
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16861700
Backup the system cause this may break it, but take one drive at a time offline (or pull one at a time out) and copy that file 345mb file.  Until you speed up disk access you are going to have aproblem.  Perhaps one of those HDs is not seated all the way or is bad, just not bad enough to be dedected as bad.
In the BIOS make sure hyperthreading is off - I'm sure you've set things to off, only to find out the setting didn't SAVE!!! How annoying!

-gsgi
0
 

Expert Comment

by:vally7
ID: 16862360
Hey gsgi,

i'd hope that pulling a drive out one at a time wouldn't break the system as surely that would negate the whole RAID 5 concept!!  But I agree that if, after upgrading the BIOS, the problem still exists then we will concentrate on checking the physical disks.  We'll double check the drives are all seated properly, and then pull the disks one by one and run the file copy test.

As for the HT, it is definately turned off as the number of processors that Windows is reporting is two, rather than four with HT turned on.

If you got anymore ideas then chuck them my way

cheers bud
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16863190
The russian roulette part is that if disk 5 say is dead, and you pull out drive 1 first, now two drives are dead and you RAID may blow up.

You said that you upgraded the BIOS on the 641?  Run the Raid diagnostics from it -- now within windows.  then do it again using the hp utility in windows?  Why twice?  Because this is weird and we are searching for inconsistencies.

The next thing is going to be to partition off a test area, and reinstall SBS 2003 on it without ANY updates, to see if it is an update that is killing you.

-gsgi
0
 

Expert Comment

by:vally7
ID: 16864195
hi gsgi,

thanks for this, but from our point of view, there's some fundamental flaws in what you're asking us to do here.  I know that none of the disks are dead, we have checked the Raid diagnostics and the HP diagnostics, but I will double check to make sure.  This system is in a live environment, so partitioning off disk space and then reinstalling SBS is not really an option as the server, whilst being unresponsive, is business critical and is working and allowing the user's to do their jobs.  In a test environment this would be fine, but unfortunately it's not.
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16866539
Are you running Symantec?

-gsgi
0
 

Expert Comment

by:vally7
ID: 16871292
nope, we were running Sophos, but have uninstalled this for now.
0
 

Expert Comment

by:vally7
ID: 16871298
By the way, I'm assuming you meant Symantec Anti-Virus when you said that! ;-)
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16871511
Yes I meant Symantec AV.  Have you called 1-800-HP-Invent?  They no doubt will take a 5 minute-to-copy-300-megs as a serious problem ...

You either have to set up identical hardware to figure this out or diagnose-and-fix the business critical system live -- and too some extent unfortunately that may require tests.  I'm with you - having identical hardware to play with is better - just more expensive.  So I need to figure out what you are comfortable doing and what you are not.

How about booting a live linux cd?  I like knoppix.  If you do that, and copy the 300 meg file, and it goes quickly, you can be more confident that the problem is a software issue related to MS.  If it is still slow, this may add evidence to a motherboard, disk controller, memory, or disk problem.

-gsgi
0
 

Expert Comment

by:vally7
ID: 16898139
Hi gsgi,

I've been away.  I think we're going to to try booting to the Linux live cd tomorrow morning.  I'll let you know how it goes.
0
 

Expert Comment

by:vally7
ID: 16915987
Hi gsgi,

Just to let you know, we got to the root of the problem.  It was caused by a sharepoint and sbsmonitoring problem we were encountering.  Anyway, thanks for your time
0
 
LVL 1

Author Comment

by:comptech_engineering
ID: 16939793
So anyway... back to our question guys. We have tried everything suggested... any ore ideas? :-)
0
 
LVL 13

Expert Comment

by:gsgi
ID: 16942305
You still have not precisely told us what 'slow' means.  How are these boxes "slower"?  If you went from mirrored to raid 5 maybe it *is* slower.  Maybe the new servers hate the switch, although you swapped out the switch.  Are they complaining about fractions of a second while apps open, printing delays, login delays, logoff delays, menu scrolling pauses, what?

-gsgi
0
 

Expert Comment

by:vally7
ID: 16992314
hey back again,

we've had reports from the users that the server has become unresponsive again.  We've only just got the call, so we're going to go in tomorrow to find out exactly what they mean.  I'll come back with more details tomorrow I guess.
0
 

Expert Comment

by:vally7
ID: 17024413
We've just noticed that our Smary Array 641 card is set to 100% Read and 0% Write on the Disk Array Cache.  Apparently without buying a Write Cache enabler battery pack you can't change this to 50% Read 50% Write.  This sounds like it could be the cause of the problem though.  Are you using RAID cards yourself?
0
 
LVL 13

Expert Comment

by:gsgi
ID: 17024469
I use a 641 and percs.  I doubt that is the issue.  -gsgi
0
 

Expert Comment

by:vally7
ID: 17071534
we added the write cache battery and after the battery charged we got some improvement in performance.  We've also stopped Sophos Puremessage from scanning the information store constantly.  Again this improved performance.  I'm sitting with the server now and there are not that many users in, but the disks appear to be thrashing away every now and then.  I'm running filemon and have noticed that the store.exe process keeps trying to access a file in c:\windows\temp and in the results column it says 'No such file'.  Anyway, I'm going to be with the server for the next 2 days trying to work out what is going on, so I will let you know if I find anything.  If you have any suggestions on things to try then please throw them my way.
thanks
0
 
LVL 13

Expert Comment

by:gsgi
ID: 17073401
0
 
LVL 13

Expert Comment

by:gsgi
ID: 17225616
paq / no refund and have netminder send an admonishment to comptech_engineering for not following up with this thread via e-mail because a lot of us worked hard on this one.  Also comptech_engineering never DEFINED slow as I requested twice.

-gsgi
0
 

Accepted Solution

by:
ee_ai_construct earned 0 total points
ID: 17296892
PAQ / No Refund
ee ai construct, community support moderator
0

Featured Post

Concerto Cloud for Software Providers & ISVs

Can Concerto Cloud Services help you focus on evolving your application offerings, while delivering the best cloud experience to your customers? From DevOps to revenue models and customer support, the answer is yes!

Learn how Concerto can help you.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

This article explains the fundamentals of industrial networking which ultimately is the backbone network which is providing communications for process devices like robots and other not so interesting stuff.
This article will help to fix the below errors for MS Exchange Server 2016 I. Certificate error "name on the security certificate is invalid or does not match the name of the site" II. Out of Office not working III. Make Internal URLs and Externa…
Michael from AdRem Software explains how to view the most utilized and worst performing nodes in your network, by accessing the Top Charts view in NetCrunch network monitor (https://www.adremsoft.com/). Top Charts is a view in which you can set seve…
Monitoring a network: how to monitor network services and why? Michael Kulchisky, MCSE, MCSA, MCP, VTSP, VSP, CCSP outlines the philosophy behind service monitoring and why a handshake validation is critical in network monitoring. Software utilized …

755 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question