garyw
asked on
Multi homed NT Server in Domain - Slow access to workgroup. Why?
I have an NT server with two network cards in it. One is attached to an NT domain of which it is a member (Just a server, not a PDC or BDC). The other card is attached to a workgroup. I am using TCP/IP only. Ping works fine to all machines in the domain and to all machines in the workgroup. There are no pauses or problems with ping when i use the ip address or the ip name. I am not running a WINS server anywhere but am using lmhosts.
The problem: When I do a "net use z: \\booth1\share", where booth1 is a machine in the workgroup, it locks up for around 50 seconds before the command is successful. Simlarly, if I browse the workgroup using "network neighbourhood", the list of machines in the workgroup (about 10) comes up in a few sconds but if I then click on one to view the shares, another 50 second gap occurs.
I ran the network monitor and got the following results. When a "net use ..." is performed on the server to a client in the workgroup, there is an immediate name query from the server which is answered straight away by the client. Some more stuff takes place within a second (a session request by the server and a positive session response by the client). Then there is a 50 second gap with a couple
of TCP packets going from server to client. After the huge pause, the server sends a C negociate message with a list of dialects (of LAN Manager?) in it. Immediatly, the client replies (it picks number 7) and everything then works.
How do I get it so that access to the workgroup machines do not have this 50 second pause? I need to keep access to the domain as well, of course.
The problem: When I do a "net use z: \\booth1\share", where booth1 is a machine in the workgroup, it locks up for around 50 seconds before the command is successful. Simlarly, if I browse the workgroup using "network neighbourhood", the list of machines in the workgroup (about 10) comes up in a few sconds but if I then click on one to view the shares, another 50 second gap occurs.
I ran the network monitor and got the following results. When a "net use ..." is performed on the server to a client in the workgroup, there is an immediate name query from the server which is answered straight away by the client. Some more stuff takes place within a second (a session request by the server and a positive session response by the client). Then there is a 50 second gap with a couple
of TCP packets going from server to client. After the huge pause, the server sends a C negociate message with a list of dialects (of LAN Manager?) in it. Immediatly, the client replies (it picks number 7) and everything then works.
How do I get it so that access to the workgroup machines do not have this 50 second pause? I need to keep access to the domain as well, of course.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
With the WINS Server you acelerate the Browsing process, because each machine instead of broadcast to search al machines it directly ask to the wins server for all the workgroups,
machines and shares.
IF there is no problem with your plans, and your have only
one workgroup try to put this the same name as the domain, this will help you because the PDC is the master browser. And only put in the lmhosts. the lines I say you in the proposed answer from me.
machines and shares.
IF there is no problem with your plans, and your have only
one workgroup try to put this the same name as the domain, this will help you because the PDC is the master browser. And only put in the lmhosts. the lines I say you in the proposed answer from me.
With the WINS Server you acelerate the Browsing process, because each machine instead of broadcast to search al machines it directly ask to the wins server for all the workgroups,
machines and shares.
IF there is no problem with your plans, and your have only
one workgroup try to put this the same name as the domain, this will help you because the PDC is the master browser. And only put in the lmhosts. the lines I say you in the proposed answer from me.
machines and shares.
IF there is no problem with your plans, and your have only
one workgroup try to put this the same name as the domain, this will help you because the PDC is the master browser. And only put in the lmhosts. the lines I say you in the proposed answer from me.
ASKER
I will move the booth1 machine from the existing workgroup to a workgroup with the same name as the domain and edit it's lmhosts as you suggest.
I will then go to a server on the domain and do a "net use z: \\booth1\temp" and the 50 second delay should be gone? It will take a day or so as the system is in heavy use at the moment (despite the bad network delays)
I will then go to a server on the domain and do a "net use z: \\booth1\temp" and the 50 second delay should be gone? It will take a day or so as the system is in heavy use at the moment (despite the bad network delays)
ASKER
As I understand it, WINS maps machine names to ip addresses. If I ping a machine in the workgroup (e.g. ping booth1) from the server in the domain (the one with the 2 network cards), it works straight away. If the problem was WINS, wouldn't this take 50 seconds as well?