I have been given a problem that goes like this.

An existing Arev client is evaluating the OI version of the application. However we have some inexplicable performance issues with the NT service installed.

Without Service:

OI takes 20 seconds to present Login/password dialog

Additional 40 seconds before the application MDI is presented.

No other users on server.

With Service:

OI takes 90 seconds to present Login/password dialog

Additional 4 minutes before the application MDI is presented.

No other users on server.

Server Configuration:

NT4 SERVICE PACK 5

Dual PII 450

512 mg Ram

Dual Token ring cards.

running IPX and tcp/ip.

Token ring frame type is set to Token ring 802.5

No Ras installed.

No Exchange installed

No Sql server.

It is the PDC.

No BDC and network is a maximum of 40 connections.

30gb drive array. ( dont know if raid or not)

Client Configuration:

Windows 98

PII 450

128 mb Ram

Olecom PCI 16mb 3137 series Token ring adapter.

Primary Network logon is microsoft networks.

and also has Novell Intranetware client 2.20 installed.

NT service is verion 1.5.

The client is an existing Arev 3.11 user on Novell w/Nlm. The Novell server is ancient 486dx 66 (uses Microchannel bus) but performs well. The network is token ring.No room on this server to host the OI application and it's not worth attempting to upgrade it, so the application is Moving to OI on the NT server. ( if we can get it to perform).

Things that I have tried.:

- Reinstalling OI and NT service. No change.

- Specifying the Frame Type - No change.

- fiddling with Revparam setting serveronly.

- The client is communicating with the NT service as it loads the tables (albeit slowly) with serveronly=true. It wont load when the nt service is stopped.

- Without the serveronly parameter set and the service stopped it loads ok, with no major performance issues.

Access rights all appear to be fine. SYtem user has correct permissions.

Both the OI application and data is hosted on a NTFS volume on the server.

Any Ideas?.

TIA

Sam McDonald

Tristalee Pty Ltd.

[email protected]

I suiggested removing additional protocols, No difference.

The client requires both IPX and TCP/IP bound on the server and clients.

Sam,

I don't suppose the NT Server has a 3D screen saver does it? If it does turn it off and try again.

Tim

Sam,

This may seem like a pain, especially with large amounts of workstations, yet it has been known to help quite often in a situation such as yours:

1)

Log out all users and stop the service from the NTService path on the server (lhsrvc stop).

Remove the service (lhsrvc remove).

Recreate the service (lhsrvc create).

2)

Unmap all of the workstations from the server.

3)

Reboot the server.

4)

Reboot the workstations.

5)

Remap the workstations to the server.

6)

Launch your application.

Again, it may be a pain but after you do this once you shouldn't need to do it again.

Good luck,

Stephen

Tim,

Thanks, but there is no screen saver enabled on the server at all.

Sam.

On any NT service configuration I have tried, I ALWAYS run into the same problem you describe. The ONLY way I have found to correct it is by REMOVING IPX….PERIOD…Once I did this the system flies after I recreate the service as described earlier. I am not sure what the Microsoft IPX protocol causes on the workstation, but the problem is on the workstation side not the server…If I use Novell's Client the problem is not as dramatic. Hope this helps…

Paul

Paul,

Thanks for your input, I've been away for a week so I have not had a chance to respond.

Tossing out IPX may not be an option for the client as they are running token ring and OI is the new boy on the block. So if it wants to stay it needs to fit in!

They are running a novell client so I would hate to see it with the microsoft client .

Anyway, I will pass on your experiences to the client and see what

happens.

thanks again

Sam

Stephen,

Thanks for the response. I would have replied sooner but I took a week off to move house.

I have been told they have done this previously with no change in result.

Anyway, I just wanted to clarify the ' Unmap the workstations from the server'.

By this I take it you mean go to the workstations and remove any drive mappings to that server.

If so I will get them to try it again.

Does anyone know or have a theory on why this does/should make a difference. The reason i ask is that I am certain the client thinks I am stalling. So any plausible explanation would be appreciated.

Also If iot doesn't have any effect, What next?.

Cheers sam.

View this thread on the Works forum...