Two PCs connected to LAN but can't see each other

hungta used Ask the Experts™
Hi all,
We had a small network around 40 PCs running Win2K, server is running WinNT 4.0 using DHCP to assign TCP/IP address to clients. We got an insolvable problem.
There are 2 PCs connected to network,it means they can access to network printer, mail server , but they can’t see each other. “Requesr time out" is the result when sitting at one PC and ping the other.
Thanks for any help
Hung Ta
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
JJ2Technical Manager

Maybe one of the PC or both have a personal firewall installed. In this case, the firewall should be modified to allow the specific IP address of the of the other machine.


Thanks JJ2,
I'm sure that having no firewall installed there. Maybe the root cause is something else. Please help.
JJ2Technical Manager

Does your network have one IP subnet only or more than one that is defined in your DHCP server?
Ensure you’re charging the right price for your IT

Do you wonder if your IT business is truly profitable or if you should raise your prices? Learn how to calculate your overhead burden using our free interactive tool and use it to determine the right price for your IT services. Start calculating Now!

You know that this questions is repeated 2 more times!  How about deleting the dups!


Hi JJ2,
Only one subnet defined in my DHCP server from to
Hi ocon827679,
I have pressed submit several times so it causes duplicate contents. I have asked expert for mixing all to only one.
Are both of the machines on the same subnet? You have 2 seperate networks listed in your DHCP definition. You either have to allow all clients to be on one subnet or, unless you want to create a superscope in your DHCP scope settings.
Oops, I meant to say either or

Either network will work, and if you have multiple protocols enabled on the client boxes, (i.e. netbeui,netbios) that are non routable, you would be able to use network resources, but not have the ability to ping (tcp/ip) each box.

We need more info.  What type of device(s) are the PCs, server and printers plugged into...I would assume you have a switch of some sort.  Do you have a router on your network?  If not, is your server set up to do IP forwarding?

What are the IP addresses of the PCs in question?  What are the subnet masks of your networks?

If you are doing no forwarding, or have no router, and your subnet mask is, devices on the will not be able to communicate via IP (ie ping) to devices on the network.  NetBIOS will still work, as long as your switch isn't bridged into seperate VLANs for the two seperate networks.

Why do you have two subnets?  If you only have 40 devices, there is probably no need to complicate that network by doing any routing, depending on your situation.



Hi Bryan,
Both connected to one 3Com Hub via UTP cable Cat5. We had one Cisco Router.
One is assgined, the other is, and subnet mask is
Neither TCP/IP nor NetBios could be realized.


Thank God! it works well now after removing and then installing network adapter at each box. But what are the root causes?

With the subnet mask you are using, both the 172.16.6.x and 172.16.7.x devices will be in the same subnet, so routing wasn't the issue, unless the subnet mask wasn't set properly on one or both of the PCs.  

Did you try doing a release/renew IP address on the PCs before you re-installed the NICs?  If you had some DHCP issues, that might have fixed the problem.



I tried many times even set new static IP address for both, but no effect.
Hung Ta

Well, Windows can be a wily beast, and sometimes just uninstalling and reinstalling the TCP/IP stack fixes problems.  Unless the problem resurfaces, you may never know what happened.  Follow up if the problem resurfaces and we can attack it again!

I'm glad to hear that you have the problem fixed Hungta, and for future reference, you need to take one of the scopes out of your DHCP server to ensure trouble-free communicaitons on your LAN, or input each machine name and IP address into the host file on each machine.

For the sake of Bdonalds, I'm inputting this information into the discussion.

The two netoworks in question are NOT on the same subnet.

Here's the breakdown in binary.

The 1st set of numbers represent: 172.16.6.XXX/
The 2nd set of numbers represents: 172.16.7.XXX/
The 3rd set of numbers represent the two networks with the subnet octect/digits in brackets.

Clearly you can see in the 3rd set of octets with the bracketed numbers (representing the actual binary digits that define the subnets in question] that these are two seperate networks. These MUST be routed if each client has received a number from DHCP representing the seperate networks, with the exception of communications on a hub.

  172   .    16  .     7  .XXX
  255   .   255  .   254  .XXX

  172   .    16  .     6  .XXX
  255   .   255  .   254  .XXX

10101100.00010000.0000011[0.XXX] 172.16.6.XXX
10101100.00010000.0000011[1.XXX] 172.16.7.XXX

The only reason that this communication would work is the fact that the machines are connected to a hub which is broadcasting packets to all ports on the hub. If this situation was involving a switch or no router [hardware/software], there would not be any communications if machines were on the seperate networks. Netbios would not work unless it is enabled in TCP/IP, because Netbios in a non-routable protocol. The only exception is that they are running on a hub, and only because the packets are broadcast to each client on the hub.

Routable Protocols Supported by Windows


Non-Routable Protocols Supported by Windows



Your comment directed at me is incorrect.  You aren't applying the concept of a subnet mask correctly.  If you are using a 23 bit  subnet mask (, only the first 23 bits of the IP address are used to determine the network number.  The remaining 9 determine the host address. In hungta's case, we have a network of, with a 23 bit mask.  

If you subnet 172.16.x.x with a 23 bit mask, your possible networks are:

Any addresses that start with 172.16.7 are actually in the network.

Here it is in binary, with pipe where the mask ends:

address: 10101100.00010000.0000011 | 0.00000001 =
netmask: 11111111.11111111.1111111 | 0.00000000 =

If you look at the address

address: 10101100.00010000.0000011 | 1.00000001

you can see that everything to the left of the pipe (the network address) is the same as  Only the host address changed.

So yes, 172.16.7.x nodes and 172.16.6.x nodes are INDEED on the same subnet when using a subnet mask, and do not need to be routed for IP connectivity.


Bryan, I stand corrected. I looked at my work again this morning and see that you exactly correct. Thanks for pointing out my error. - Paul
This old question needs to be finalized -- accept an answer, split points, or get a refund.  For information on your options, please click here-> http:/help/closing.jsp#1 
Post your closing recommendations!  No comment means you don't care.
PAQed, with points refunded (50)

EE Moderator

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial