Network Issue (advanced)

[Longy]

Background:

We are currently using an Custom Application that was hosted on our server (quote/booking/ordering system) but for several reasons we had to take it off our server and put it on a workstation with XP Pro.
We are using SQL Express on this XP machine to run the application.

Problem:

Everyone connects to this application with no problems and can use it - but randomly all users seem to get a dropped connection and get booted at the same time (no pattern to the timing, also it happens when they are using it so its not a idle thing).

The people who programme the application say its not them? but although we seem to loose the connection to app, the actual connection to the xp machine is fine (no noticeable drop out) and seems to stay connected to the share.

Question:

Is it my network or the custom app?

Is it a SQL Express setting that kicking/dropping connection?

Does XP have a timeout settings for connections? I know there is a limit of how many people can connect to a XP machine but this is happening with only 1 or 2 people connected.

It never seemed to be a problem on windows server 2003.

If it is the app at fault I need to prove it.

Things tried:
I know it could be a million and one things but I got a feeling XP is timing out the connections and dropping them even though I can connect to the shares on the XP machine all day and stay connected.

I've pinged the XP machine all day and found no packet lost.

All connections seem ok.

I've looked at the power saving options on the XP.

Any thoughts? and thanks for reading

 

[Daz]

XP has a TCP connections limit - is this being reached?
It should be logged within the System log from memory.

Sounds like it's because it's XP, especially as it didn't happen on 2003. You need to get a Server OS to host it really

 

[The Hoff]

Developing the above idea, does this only happen when a certain number of users are logged in? Can a single user stay on overnight?

Post SP2 limits TCP connections to 50, but you can increase that limit with a hacked TCPIP.sys file, and registry changes, it's well documented.

Personally I'd try and keep one/few users logged on on your PC and set a timer, see if it happens after a set period of time. I'd also then start trying to replicate it, and after finding the breaking point begin looking at Regmon and Ethereal logs.

 

[Longy]

Well I've only had only 2 users use the software for past week and they still get it randomly (no one else connecting to the machine at all - and the machine doing nothing but sql express and the app running).

I've checked the system logs and everything looks ok - nothing unusual. The XP machine does have symantec endpoints on it but I've disabled it and still get the same issue.

Its so random - 1 user it has happened 6 times today! arrghh sometimes it doesn't happen at all.

I got a feeling its when they do something in the software- but I gotta prove it and show its not my network.

Trying to replicate the problem....

Cheers for the help guys.

*It seems both users get kicked same time - need to verify that.

 

[The Hoff]

Right, I'd suggest running some monitoring software and see if that detects anything while you're troubleshooting it. I'll have a look around for some answers.

What programs/db's you running on it?

 

[The Hoff]

Been having a little sniff around, and spoke to one of the DBA's here and he suggested it maybe an ODBC issue. I'd also advise looking at the below link as I too have had issues with AutoClose before.

http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=982817&SiteID=1

 

[Longy]

Cheers Hoff - I'll look into it.

The software is a programme called Print MIS - the programmers are currently programming it to our needs so they keep updating the software constantly. It uses SQL Express.

The application just throws up a connection failure error then has to be force quit. When u next try and log in with it it gives another error "selece a valid for login" then you have to quit again then you can log in next time - seems crazy to me for it to be a network connection problem.

 

[Longy]

I managed to get the proper error - before it gives me a connection error:

[DBNETLIB][connectionread(Recv()).] General Network error. Check your network documentation.

 

[The Hoff]

Longy can you clarify exactly what it says please when it dumps people out of it?

Is there any way you could hook it upto MSDE/SQL2000 and trial it also?

*Edit*

Ok, does it give any hex code or error code along side it?
That from Event Viewer?

 

[Longy]

Thats the pop up it gives in the app - I can't find anywhere where it dumps the log file for it.

 

[The Hoff]

Googled that error and it came back with a shed load of stuff:

This seems quite interesting, it points to firewall issues:
http://forums.asp.net/t/918011.aspx

Timeouts:
http://www.experts-exchange.com/Databases/Microsoft_SQL_Server/Q_21065653.html

Googles many articles on it:
http://www.google.co.uk/search?hl=e...+documentation.&btnG=Google+Search&meta=&aq=f

 

[Longy]

Yeah been looking through google aswell, I've had the firewalls off for testing. I've even moved the box to the switch room just to make sure its not a bad cable in the wall or anything.

I think its gotta be down to either a)SQL Express (some time-out settings on there or b)Their software and its connection to SQL.

As I'm a noob at SQL I can't really mess with SQL Express to tweak it - I'll get them to do it and pass on your advice.

Cheers m8

 

[gtruk]

Sounds very much like SQL express is the issue here. Whilst SQL express might have been stable on 2003 platform, it desotn necessarily mean it will be on XP.

I doubt very much its a problem with the network. Its either a problem with SQL timing out for whatever reasons, or a mix of the software and SQL on XP.

Ive had similar issues on MySQL. See if there are any updates for your version of SQL express.

 

[The Hoff]

SQLE works fine on XP, period.

 

[Longy]

Well although the firewall was disabled (in symantec endpoints) and looking through the websites about the error and firewall being mentioned a few times I thought I would complete remove symantec just in case it still had hidden things running which symantec likes to do.

So since moving the box and removing endpoints completely I've so far (touch wood) not had the problem!

I've never had a problem with symantec anti-virus business edition before but this new update (Endpoints) which included the crap u get like in Norton (firewalls etc) is a nightmare. Bloated and problematic and if this was the problem it wouldn't surprise me.

 

[The Hoff]

Hurrah