just a hub i'm afraid, there is not enough network traffic/users to warrant a switch!
edited for my fantakka spelling

[Edited by ragnarock2 - 7/2/2003 5:01:18 PM]

 

Hi,
We have a customer that has been experiencing network problems recently. We provide a MIS for label printers, and as part of that provide the hardware for the system etc.
Our product runs via telnet sessions and the problem is thus:

The customer has a W2k network (installed by us), the app runs on w2K and the clients connect in via telnet. The hub is a 3com superstack, 100BaseT item, and all cabling was done by BT before we got involved

At the same time of the day (on random days), the telnet session will slow right down, and kick random users off the telnet session, I have looked through the W2K logs, and also Unidata logs (what our software runs on) but there are no problems reported.

My best guess is that this is an environmental issue (i.e. some power surge due to something being turned on) and/or various network cables are dragged over powercables etc, or they are experiencing power surges/spikes etc (the hub is not protected by UPS)

Any other thoughts on what it can be? Any thoughts on how we can prove it is the power etc? As the customer is munching my nuts but I am at a loss as to where the problem is!

 

are you running telnet over a wan? what speed are the links? what routers are you using?

Si

 

I think you've told us the cause already, you mentioned BT

 

There is no WAN, just a small LAN (only 8 Users) so as such, no routers either!

 

Hub or switch?

 

That's okay - sometimes find the autosensing on switches can cause problems if it doesn't set the duplex properly.

Have you got Perfmon running and logging to a file? It might show a CPU spike or something. Alternatively, if nothing shows up you can look away from the server.

 

Nothing doing in perfmon - the server is behaving fine - that's why I am veering toward some environmental issue (that and the fact that the network is fine all other times of the day!)

 

Same time of day may give a clue.

Any chance you can advance the clocks on all machines involved by say six hours until it happens again ?

This way, if it occurs at the same time as it does now according to the machines, then it is something on the machines. If it occurs but at the same wall clock time as it does now, then it is external and that would narrow it down for you.

Hope that makes sense.

Cheers

Ian

 

Not go no virus scans or audits or anything scheduled have they?

 

good idea RE clocks, but as it is on random machines, I would guess it is not a problem with the machines themselves. There is nothing scheduled to run at that time of day (all back-ups, checks etc are scheduled for overnight)

 

You don't have some idiot sending large e-mails to 20 of his mates several times a day, do you?

 

Nope, nothing like that either!!!
The only thing that we originally thought could be the problem, was that there old system was a lotus approach file (of 1.5GB) and they may have been opening/sorting this over the networ, but we have been assured they havent touched the file (mind you, who ever believes end-users?)

 

ANy more thoughts?