Use of UNC networkmappings between 2 servers takes minutes.

[rwell]

Hello,

I'm kinda stuck with a strange situation, and after reading loads of websites and trying / checking everything I could comeup with, I'm hoping some of you might be able to give some advise.


Here's the issue:

NOTE: All servers are Windows 2003 with the latest updates installed.

Two servers are configured as Terminalservers (called S03 and S06).
One server is acting as fileserver (called S05).

Last week two (3COM) switches brokedown, and I replaced them with HP ProCurve 26xx´s. These can be managed, but this will be done lateron so they´re running unmanaged for now. According to the Procurve´s manual they won´t interfere with networktraffic by default.

When I logon to S06, and openup a explorer and type \\S05\ a list with available shares would normally be shown. Now it stalls for several minutes. Network mappings created by the logonscript are also this slow. When the mappings are finally browsable, it stays fast until you stop using the share for, say 15 minutes.

This is also the issue on S03, but there the delay is only several seconds.


I´ve checked the following:

1. - Isolate the problem.
When I try to browse the shares of S05 from any other server, laptop or workstation, the shares are shown instantly - as they used to have.

2. - DNS issues.
Servers resolve on all servers using ping. ipconfig / all show same DNS and WINS settings on all servers.

3. - Eventlog.
Eventlog is silent. No errors.

4. - Connectivity.
Servers are on the same switch. IP traffic goes fast, no problems found.
netstat -e shows no interface errors on any NIC.
Also, the Compaq NIC software on S05 shows no errors.
When I map shares using IP´s, access is granted immediately.

I´m sure I´m forgetting some things i´ve tried, but this covers it quite well I think.

I hope some of you might point something out that I didn´t think of.


Thank for reading.

 

[Freestone]

With the introduction of new switches, is it now possible that the NIC on S06 (and perhaps S03) is strapped for say 100MB full-duplex whereas the new HP switch ports are set for auto-negotiate?

 

[rwell]

Thanks for you reply.

The NIC´s are configured to auto negotiate. They´re connected to 100MB ports on the new switches, and also operating on this speed in full-duplex.

Exept S05. This server has it´s networkcards teamed using Compaq software, and is running at 200MB full-duplex.

 

[rwell]

At night I figured the problem might be caused by the teaming of the two NICs. When I simply disconnected 1 UTP cable this morning the problems where gone.

The reason that S03 was connecting faster to the shares than S06 might be caused by the 'HP Network Configuration Utility' software, wich is installed (and in use) on S05 and S06, and isn't on S03.

Sadly enough I can't spend any more time investigating further on tis matter at this point. (can't believe that products of the same brand give this kind of problems where the situation with 3COM 22xx switches worked just fine).

Perhaps in some weekend I'll try setting up a new teaming to see if I can get things to work.


Thanks for reading.

 

[wdoellefeld]

Sounds as if you didn't setup the switch to support the team. The switch needs the aggregation on the ports so the both the switch and server can speak LACP.

(yay! shameless advertising. my side business)

http://www.frcp.net