Link to home
Create AccountLog in
Avatar of Stooze
Stooze

asked on

Connectivity issues on Windows 2003 Domain

Windows 2003 R2 domain and we've just added a 10 new pc's to the domain. We're running a SQL database off the server with application local on desktops. All XP SP2 desktops, they all log onto the domain using same generic login.

A couple of weeks ago, just after the pc's were added and some new cabling was installed, the app. connected to sql database started to show blank pages. It's only happening on a few of the new installed PC's and if you log out of the app and log back in, it's OK. There's another SQL app runs without error and Sophos.

The people who support the app say it's a networking issue.
They say a micro network drop out would not have a huge impact on the functionality of database and checking is done at Windows file sharing level. However if the connection to the server was lost for a millisecond and reestablished at Windows level, the application would not automatically refresh the links and have to be restarted. Stabilize your network this is not our issue.

I had some issues with the cabling and got the guys to re-check, they say it's OK.
The network is served by a really old 3com switch, all the rest of the hardware is new.
Avatar of slam69
slam69
Flag of United Kingdom of Great Britain and Northern Ireland image

Could the increased volume of traffic due to the new PCS be causing a drop in  packet speed queuing at your switch? IF THE WHOE NETWORK IS RUNNING THROUGH THe ONE SWITCH COULD HAPPEN AND MIGHT BE TIME TO REPLACE THE SWITCH?

Am i right the 10 pcs are additonal to the existing network?

sorry about teh caps
Avatar of Stooze
Stooze

ASKER

Hey slam!

cheers for your input!

the switch definitely should be replaced but I'm not convinced it's the issue.
There's only 12 pc's on the domain and 9 at a second office with their own bdc that ts into the sql app on the server over vpn link, no issues at other office or with anything else obvious on network.
Avatar of Stooze

ASKER

one difference is that the pc's never used to log on to domain, only local but surely would affect all machines not just a few
so the domain structure is new then? are you using one server to act as your gateway and your active directory and dns?

could be the traffic through that server is causing the bottleneck, I think some packet speed tests could help you here have a look at your network traffic.

obviously if everything was local before you didnt have teh issues with network authentication being required whereby teh sql app has to check with ad the user account is ok ( thats of course if you are using windows authentication for your sql app), however if you are using sql quthentication whereby they enter a seperate username and password ignore this last bit
Avatar of Stooze

ASKER

Structure is the same both servers are DC's with AD and both DNS, DHCP. So really I've just replaced 10 pc's and made all 20 of them authenticate using windows standard AD logons they all used to log on locally. This has been done in two implementations and no errors till last one. Cisco 1800's are gateways, server runs RRAS to authenticate external vpn users, All desktops function properly bar 3 but they are all new. Sql authentication is automaticlly handled by sql instance.
I was wondering if there was latency caused by windows authentication after pc's have logged on?
As it's inconsistent and not happening on all of them, it seems unlikely.
New cabling is high priority to evaluate as this also coincided with issue occurring, I guess the switch could be the issue as all other hardware is less than 12 months old.
:)
ASKER CERTIFIED SOLUTION
Avatar of slam69
slam69
Flag of United Kingdom of Great Britain and Northern Ireland image

Link to home
membership
Create a free account to see this answer
Signing up is free and takes 30 seconds. No credit card required.
See answer
Avatar of Stooze

ASKER

slam69 you rock! thanks for your input!
I've got a spare switch to try too.

I do however think that it could be another issue...there was some windows update recently with Chimney that pushed some cpu time onto the NIC, if the NIC didn't cope with this creates a myriad of issues. I've got the fix I'm going to try on Friday.
Failing that I'm back to cabling people. I'll let you know if I need you to come out! LOL :)
haha cool be worth trying a different NIC as well then also check your configurat0ion on your NIC if you change the switch might change the neogtiation speed, auto negotiate might be the way forward if your not sure but you might need to tweak manually to get the best performance
as for tha call out how much do i charge from england to OZ haha

let me know how it goes and if we need to do anythgng further
Avatar of Stooze

ASKER

OK,
I've disabled chimney on the server and reset winsock on affected clients to no avail.
There are some cabling issues, ie to wall ports are patched to the same socket in patch panel but I've isolated these. Server and client machines are patched.
The interesting thing is this is intermittent.
Additionally I have now been informed that it has happened infrequently before.
The app that connects to the sql database is a clunky VB thing that does have issues...aaarrggh!
So I guess it's back to the weakest link...the switch.
I'll pop in another and see if it resolves, failing that I guess I'll have to test the flow and if that's ok it's got to be the app?
:)
ik think thats th eonly route to go down, would have been nice for the to have let you know before that this used to happen but then thats the mindset of the end user :O)

Im still banking on the switch
Avatar of Stooze

ASKER

hmnn didn't work...
Do you think maybe it's just the app is dodgey and now it has more network traffic it's not coping?
I suppose it could be that you could try a network monitoring tool to see if there is any packet delay in traffic across your LAN but if that was the reason, only way to be testing that is to try running it locally or just between 2 machines in a test environment to see if the problem is removed

turning into a proper head scratcher this!!
Avatar of Stooze

ASKER

Awright Geeze?
:) Well it was a bunch of PC's at the sattelite office authenticating across a tenous WAN link.
I tested the latency on each router and it was avg 300ms on Sattelite office. It's now 93avg on Sattelite and 230 on main office. Seems to have worked.
cheers for all your input
I would have thought selecting a combination of answers as opposed to deleting would be better as quite a lot of useful stuff in here to try
Avatar of Stooze

ASKER

I agree!
excellent, nice to see when someone has a rethink :O)

Was fun working thorugh this one and glad you pinpointed the issue in the end

Speak to you again sometime

Jay