Intermittant network trouble

[Ken Flickinger]

LC2000 netserver; win 2003 SBS; 3 36 gig scsi drives in raid 5; intel pro1000 NIC; gig of ram

trouble shooting so far;
replace gigabit switch between Terminal server and data server(lc2000); replaced the NIc in lc2000; updated the drivers for the nics.

I have about 12 Terminal server clients that access this server for accounting purposes, AccPac to be exact, I have a 3 FAT clients as well they access the data on this server conventionally. Both types of clients are having data transfer problems at the same time. That is how I ended up in the LC2000 box. It is common to both types of clients. This setup has been running for about 9 months. Only recently have these issues been so prevalent and they are getting worse. There are no errors being reported of any significance by the event viewer. The only evidence I have that there is a communication problem is that outlook runs on the terminal server and exchange runs on the lc2000 and the application event viewer shows that outlook lost contact with exchange and then regained contact again within the same minute.

If my accounting users are doing something they lose what ever they were working on because of coruption. The AccPac application say "oops" and freezes. Poof there goes your last 10 minutes of work. Multiply that by 12 and imagine my popularity.

Other users in AccPac who did not touch any buttons carry on none the wiser.

There are no predictors. The events occur when its idle, when its busy, One user (any user) or 20 users, whatever. They hit they're gone, we just keep working because everything works imediately afterwards we don't know if we'll get 30 seconds or 3 hrs of service till the next one. We are not happy.

No amount of reboots or service packing or up dating has any effect.

I have the ethereal network analyzer but unless I happen to be capturing data at the time one of these events occurs, it tells me nothing.

I am going to replace the server. I pray that that is the end of it. If any of you can help me resolve this maybe I can still be a hero

Thanks
wits end (3 weeks of tracking this problem to its most expensive end) ;-(
Ken

[Mic V.]

Ken,

First, welcome to the forums, and let me add my condolences. I know the feeling.

Your description is excellent, but if I read your post correctly, I think you may want to re-post this to another forum (Microsoft, networking, or servers) -- this is the forum for HP UNIX (HP-UX).

Regards,
Mic

[Ron Kinner]

Some switches will turn off a port that has too many errors. Others will turn it off and then back on. This could be what you are seeing. Can you check the switch for errors or resets? What switch(es)? Can also be a software error on the switch that causes it to reboot. Should be something in its log files.

I have also seen a worm/virus periodically overload the network which keeps everyone else from getting through. Since you say this just started a few weeks ago this could be the problem.

A "broadcast storm" usually caused by a loop can do the same thing.

Ron

[Mohanasundaram_1]

Hi Ken,

As suggested already your switch may be resetting. Get that checked. This can also be confirmed if all other servers connected to the switch experiences the same problem.

If not, I would strongly suspect a duplicate IP address problem. In HP-UX I would have told you how to check, but am not sure here.

I remember this duplicate IP address problem does not provide any clue but takes a lot of your time. The problem also matches your symptom.

If duplicate address is the issue, I am afraid it will still haunt you after changing the server.

I would not change the server until I have isolated the problem area. It can be network/OS/System hardware. So isolate it.

With regards,
Mohan.

[Ken Flickinger]

I took the first responce to heart and posted my trouble in the windows area. I thank you for all your suggestions. I think the last fellow had it right(ish) since I did replace the server to no avail. But the problem was not IP addresses. It was I believe the the checksum settings on the Nic they were turned on after having those settings turn off no more errors.

None so far, 3 days now.

Again thanks for your suggestions

Ken flickinger