Topic: Problem pinging machines on the same LAN

[mscdex]

I'm running SME7.3 as server and gateway and everything. I have two machines plus the server on a single switch as a small LAN. On both client machines I can ping and access the server (via browser and My Network Places) just fine. However, one either client machine, I am unable to ping or connect to the other client machine. When I try and ping the other it comes up with "Request timed out" but it did resolve the hostname correctly. On the server I can ping both client machines by IP just fine (but not by hostname, that's a separate problem I suppose).

I have both client machines set up the same as far as the LAN adapters and protocols being used are concerned. Changing the NetBIOS settings around on both have not worked either, as well as trying to disable File and Print Sharing service under the LAN adapter properties.

I've spent several hours googling, pulling hair out, and trying everything I can think of but nothing thus far has solved this issue. Searching on the forum here didn't lead to any solutions, most of the results I came up with were related to VPN situations.

EDIT: I'm not sure if it's a subnet problem or not exactly but, the one client's IP is 192.168.14.254 and the other's is 192.168.13.255. Both have subnet set to 255.255.240.0. The server's IP is 192.168.1.1.

EDIT 2: Both client machines are running Windows 2000 Professional with SP4 installed and neither have any kind of firewall or antivirus or similar security software installed.

[brianr]

Sounds like a firewall problem on one or both of the client machines.  Do you have somes "internet security" package installed (Macafee, Norton, Avg,etc)? If so, disable the firewall and try again.

[mscdex]

Sounds like a firewall problem on one or both of the client machines.  Do you have somes "internet security" package installed (Macafee, Norton, Avg,etc)? If so, disable the firewall and try again.

No, the two machines don't currently have any firewall or antivirus or anything like that installed. I forgot to mention they are running Windows 2000 Pro with SP4, so there is no built-in firewall to deal with also. This one really has me stumped

[brianr]

I suspect your subnet masks then, although the numbers as you post them seem to work (although my hex is a bit rusty).  Check that you _do_ have the subnet masks (and the IPs?) set as you indicated.

Another idea is to change the clients to use 255.255.255.0 and change them to be in the 192.168.1.0/24 subnet.  Even just as a test.. 

[cactus]

I suspect your subnet masks then, although the numbers as you post them seem to work (although my hex is a bit rusty).  Check that you _do_ have the subnet masks (and the IPs?) set as you indicated.

The subnet mask should allow for those ip's to be seen, as a subnet mask of 255.255.240 is the same as /20, which in this case would allow for ip's from: 192.168.0.1 to 192.168.15.254.

Are you pinging on hostname or on ip? If by ip works and by hostname doesn't check you so not have any stale entries lying around in your hosts files (somewhere in system32/etc/drivers/ I believe).

If by ip is not working I suggest ip/subnet errors in your configuration.

Just noticed one of your servers is at .255, what if you change it's ip to something lower. Ip's ending on 255 are sometimes seen as broadcast addresses (to address whole network subnets). The ip number might conflict here.

[mscdex]

The subnet mask should allow for those ip's to be seen, as a subnet mask of 255.255.240 is the same as /20, which in this case would allow for ip's from: 192.168.0.1 to 192.168.15.254.

Are you pinging on hostname or on ip? If by ip works and by hostname doesn't check you so not have any stale entries lying around in your hosts files (somewhere in system32/etc/drivers/ I believe).

If by ip is not working I suggest ip/subnet errors in your configuration.

Just noticed one of your servers is at .255, what if you change it's ip to something lower. Ip's ending on 255 are sometimes seen as broadcast addresses (to address whole network subnets). The ip number might conflict here.

Neither hostname or ip pinging result in success. I saw the .255 and thought the same thing, but I wasn't sure why SME would dish out such an IP if it wasn't valid? I will try using a 255.255.255.0 netmask and see how that goes tomorrow.

[brianr]

I saw the .255 and thought the same thing, but I wasn't sure why SME would dish out such an IP if it wasn't valid?

I agree that the ip with .255 must be wrong, Check what range you specify(ied) in the console configuration, perhaps the DHCP task does not check it, and perhaps the console config task allows it through?

[CharlieBrady]

I agree that the ip with .255 must be wrong...

No, only 192.168.15.255 is a broadcast address in the range which is given.

Do you really have anything close to 4000 nodes on your network? If you only have a few hundred, consider adding a router or two and partition your network into /24 nets.

Check what ARP is doing on those two clients.

This looks like a client OS or config problem, to me.

[brianr]

No, only 192.168.15.255 is a broadcast address in the range which is given.

Charlie, I don't understand your comment - is is reasonable to have the IP 192.168.15.255 assigned to a server?

[CharlieBrady]

Charlie, I don't understand your comment - is is reasonable to have the IP 192.168.15.255 assigned to a server?

No, but 192.168.13.255 is OK, if 255.255.240.0 is the netmask. The network address is 192.168.1.0, the broadcast address is 192.168.15.255, and all other IPs in the range are valid host addresses.

That's not to say that some people and some software might be confused by such a network. In particular, it's crucial that all nodes on the network are configured with the same netmask.

[brianr]

yes I see, thanks for that clarification.  As I only ever use 255.255.255.0 it had never occurred to me!

[mscdex]

Just tried 255.255.255.0 and the machines can now ping each other. We most likely won't need 4k addresses, I was just putting something in there temporarily that would be more than what we need. The current location where I am setting all this up at, they don't have any routers that I am aware of and I have no control over that sort of thing. What are possible reasons as to why other netmasks do not work? 255.255.255.0 would allow 253 client machines (.0 and .255 for broadcast and .1 for SME server), but it is possible they may install additional machines that may very well go over this limit (computer labs, etc). Is the problem here that 192.168.13.x cannot communicate with 192.168.14.x for some reason or is it something else entirely?

Thanks all for your insight.

[brianr]

I still think that the IP address 192.168.13.255 might be the problem, as Charlie says

That's not to say that some people and some software might be confused by such a network. In particular, it's crucial that all nodes on the network are configured with the same netmask.

[mscdex]

I still think that the IP address 192.168.13.255 might be the problem, as Charlie says

Both machines had the netmasks set to 255.255.240.0 from the DHCP server, according to ipconfig. They are the only two on the network, besides the SME server (which also has the same netmask of course). I'm thinking now that I tried to allocate more hosts than a class C allows (254 at most) and used a netmask that was not valid for class C. Perhaps that was the problem and that using a class B address will allow me to supply the number of hosts I was originally trying to use.

EDIT: Yup, just as I suspected. I didn't think about the class C host limit. I'm now using a class B address and netmask and everything works fine now. Thanks again all for the help! 

[william_syd]

EDIT: Yup, just as I suspected. I didn't think about the class C host limit.

While strictly Class C does have a host limit, altering the subnet mask and moving into classless addressing should work with the IPs/Mask that was given out by the SME DHCP server.

To accept your solution would imply that SME accepts erroneous information during set-up and from that produces an unworkable addressing assignment.

What type of switch are you using?

[mscdex]

While strictly Class C does have a host limit, altering the subnet mask and moving into classless addressing should work with the IPs/Mask that was given out by the SME DHCP server.

To accept your solution would imply that SME accepts erroneous information during set-up and from that produces an unworkable addressing assignment.

What type of switch are you using?

It's just a regular SMC 10/100 8-port (or more, can't remember off the top of my head) switch used in all the classrooms and other places.