Server/Network problem

[csgarcia]

Hello,

I have a problem with my Windows 2000 Server. My server is a Domain Controller, DNS Server, SQL Server Std Ed. and File Server. I have approx. 70 users logged in to this server. It has Dual Pentium Xeon Processors that run 2.4GHz. Everything was fine until 2 weeks ago. So I upgraded my server from 1G to 2G of memory. Im still experiencing slowness accessing files over the network. Im not sure where the problem resides, but i know its in my Server. I've sniffed my network and nothing is clogging up the network. I did notice that my SQL database file is up to 16G. Everyone is using a Front End Software to access SQL Server. But im not sure why my server is still lagging. Do i need to still upgrade my memory? Is there something wrong with my SQL Server?? I dont know?? Please help.

[Agent Smith]

Good evening csgarcia, and welcome to the Forum.

I am almost 100% certain that this will be a problem with your SQL. It seems to be quite a common issue. I have to admit, that I am not an expert in SQL type issues, but I think what you are going to need to do, is start using perfmon (or if you can afford it an aftermarket alternative) to start a rigorous process of monitoring what is actually happening.

I suspect that it is a case of the SQL server not releasing the SID's of the client machines, thus builiding up the memory usage untill you slow to a crawl. I will need to wait untill tomorrow to try and speak to a guy at work who knows more about this than me, but he may not be in untill monday.

In the mean time, I don't know if the following links may help at all.

DBForums

SQL Server Worlwide Helpgroup

SQL Server FAQ's

Your SQL stuff isn't written by Capita by any chance is it?

Hope this helps in the short term.

Regards. Mr. Mouse.

[Agent Smith]

csgarcia, I have been giving this some more thought, and I have to apologise for jumping straight on the bandwagon and blaming the most likely culprit (SQL), especially as you hinted that you think it is the problem.

I appreciate that some of the following may seem really trivial and basic, and it is not my intention to cause offence, but sometimes the really basic problems get overlooked, because of exactly what they are.

I still think that your SQL server is going to turn out to be the problem, but we should not rule out other possibilities. So, you say this has only started in the last few weeks. Has anything on your network, no matter how trivial it may seem been changed recently. Any new software added, updates and patches applied, new users created etc. ?

I have to assume that this is a business critical server, but can you, on the basis of having to troubleshoot the problem, kill any services for short periods of time, to see if they have an impact.

Do you have any other servers in your domain? And if so, what is their role? What about backups, are they handled by this box, and if so, are they running correctly (I have seen a failed backup cause all sorts of problems on a server!).

Have all your volumes got sufficient free space? What Os's are you running on the clients? What protocols as well?

I know you have done a packet capture, but with the current deluge of virus's, is it possible you have some infected machines causing you a problem? Where did you "sniff" the network, bearing in mind that if you are switched you will only capture broadcasts or local traffic? (Unless using expensive equipment) You do not mention internet access, but I assume you are running some form of connection, probably through a firewall (ISA perhaps), have you checked the traffic into that?

Perhaps a big clue to an SQL issue, is to reboot the server. If all is well for the first couple of hours, with a gradual decline in service as your database starts to increase in size, then it would seriously give me reason to place the blame there.

As I say, sorry if some of these seem trivial, and I suspect that you have already checked most and probably many more, but I need to ask to make sure.

Please keep us updated with your progress.

Good Luck. Mr. M.

[andsome]

Welcome to the forum. Hang around and contribute. I hope that our resident rodent, or someone else can be of assistance.

[csgarcia]

Hi Mr. Invincible Mouse,

I have no changes in the server, no software add-ons, no patches but maybe some windows updates and 5-7 user created. I Have one server in the domain but pretty much does nothing. The other server also has SQL Server on it pretty much just for backup. Nobody is accessing the additional server except 4 people (IT Guys). My server has 2G of Mem, a total of 140G HD space, 70G are used, I have all PC's connected to HUBS and about 5PCs and printers to a Switch. When you mentioned about Perfmon, I get this a lot when I go to Application in the event Viewer.

Event Type: Error

Event Source: Perflib

Event Category: None

Event ID: 1010

Date: 3/12/2004

Time: 11:10:05 AM

User: N/A

Computer: PRODUCTION

Description:

The Collect Procedure for the "Tcpip" service in DLL "C:\WINNT\system32\Perfctrs.dll" generated an exception or returned an invalid status. Performance data returned by counter DLL will be not be returned in Perf Data Block. Exception or status code returned is data DWORD 0.

Data:

0000: 3d 01 00 00 =...

Maybe this is causing the problem??

Thanks a million!!

[Agent Smith]

Hello again csgarcia.

Right, first I thought "Oh No!! a damn event code, nobody can troubleshoot those properly, not even microsoft, it's only a performance monitor, that can't be having an effect", ....but i decided to do a bit of surfing just to make sure.

Now I'm not so sure. I have found quite a few references to this issue, and it's amazing how many of them seem to relate to SQL Servers :(

Microsoft themselves seem to think it may be a problem with the dll and certain types of NIC KB Article 819716

But some other sites seem to think there is a work around to try before using the M$ beta fix One Example (I know this relates to NT4, but seems valid)

I have learnt more about performance counters in the last hour, than I thought I would ever want to know, so here are a few examples of sites with information you may find usefull/interesting.

EventID.net (Plenty of links on this page)

Another Support Page

Event ID's (admittedly for Terminal Server, but they seem similar to 2K)

Also, there seems to be an issue with this problem relating to Compaq (Now HP) servers using Compaq's Insight Manager, so in case that is what you have, here are some links for this.

Compaq and KB Article 186713 and even if you are not using Compaq kit, this site Chicagotechnet seems to think that the data covered in the previous two links is the cure for the issue anyway :huh:

Anyway, I think there is enough to be going on with at the moment, so I shall leave it like this for the time being.

One last thing though, one of the guys in my office today, pointed out that you should use the SQL maintenance wizard to run each night (assuming the office is empty), to clear out all the opened/generated tables and connections that have accumulated during the day. Basically, the advice was to dump the database every night, and start afresh each morning.

Please keep me posted.

Regards. Mr. M.