Dear Experts,

I did network support for >10 years so hopefully am not being stupid. But I am getting desperate as I've run out of ideas. I will try to summarise quickly. I have a Belkin Wireless Pre-N router with 4 machines connected to it. Presently all connected to the wired hub but I want to convert all clients to wireless and leave my server wired.

So if I take my first client machine, disable the wired card, enable the wireless card and reboot, it can see all the other client machines using, for instance, windows explorer. It can also access the server machine for any services that just use basic IP networking. e.g. the server's Apache service and Lotus Notes service both respond fine. But in Windows explorer I cannot access the file server share on the server any longer. And believe me I have tussled and tinkered for a long time trying to get it to work. Since I think the file server share is using NETBIOS, I think it is a NETBIOS problem. And since I can still access the shares on the other machines, it would seem to be a NETBIOS problem on the server machine only.

At first I thought it was a problem with the Belkin hub. But since the problem only happens with the server, it can't be I suppose.

Here are some things I have tried:
The IP settings on the server are identical to those of a client for which I can connect to the windows share.

The NETBIOS settings have "default" selected.

Neither server nor a client to which I can connect have an LMHOSTS file.

I have rebooted the server, the client and the belkin hub after making the change to the wireless adaptor.

I have done nbtstat -R on both machines.

However, there is a possible clue: if I do NBTSTAT on the machine that has just changed to wireless, I get 2 entries for the server (KINGPIN) and only 1 for a client (PUNG) that I can connect to fine: Here is the listing:

C:\Documents and Settings\jmasters>nbtstat -c

Wireless Network Connection:
Node IpAddress: [192.168.2.8] Scope Id: []

NetBIOS Remote Cache Name Table

Name Type Host Address Life [sec]
------
PUNG <00> UNIQUE 192.168.2.5 107
KINGPIN <20> UNIQUE 192.168.2.3 587
KINGPIN <00> UNIQUE 192.168.2.3 155
KONG <03> UNIQUE 192.168.2.2 -1
KONG <00> UNIQUE 192.168.2.2 -1
KONG <20> UNIQUE 192.168.2.2 -1

This is reaching the limit of my knowledge and I don't understand what the <20> and <00> mean. Is this relevant?

Any other ideas, solutions or advice would be greatly appreciated....

thanks,

James Masters.

If your server is running a server OS I'd install the WINS services and configure your clients to point to the WINS server. If not, use a hosts file on each of the pc's and that should fix your problem. NETBIOS uses broadcast traffic to advertise itself to the other pc's. Evidently the broadcasts arent reaching all of the PC's. Switches and hubs pass broadcasts, but routers don't. However, in this case, you are not using the router to pass traffic, you are using the router's built in switch or hub to pass the traffic. So the broadcasts should be hitting all the pc's. In my experience the automatic netbios broadcasts have never really worked correctly, which is why WINS was created. (besides the fact that routers dont pass broadcast traffic). Also, the netbios codes you see when running nbtstat can be found on google. <20> designates that the host is running a file server. <00> designates the pc is using Micrsoft Client for Networking service. <03> designates the pc has the Messenger service enabled (the service that can receive "net send" broadcast messages).

So if you use a hosts file or a WINS server, you'll bypass the netbios broadcasts all together and always have connectivity to those servers.

-------------
Firecodex
CCNA/MCSA/MCP/A+/Net+
---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

I don't think its hardware or wireless if all the other pcs are accessable and so is the server except for a share.

Hosts files and WINS are not recommendations I would make. Host files don't populate the local netbios cache. Only a lmhosts file prepopulates the netbios cache. Great for resolving name resolution broadcasts without ever hitting the wire [resolved from local cache]. You have to properly setup/import a lmhosts file to have this work correctly.

You don't mention the flavor of server [W2K/2003] but we know it can't be XP since you are running Lotus Notes.

Setting the hardware aside, if there are name resolution problems, this is not a netbios issue but a DNS server config issue. This is, of course, assuming you have not disabled netbios in bindings and its enabled in tcp/ip properties. If you see computers in Network Places netbios is working fine.

The usual mistake in configuring DNS is not having the ISP's DNS servers listed in the FORWARDERS TAB and/or workstations tcp/ip settings are pointed to the isps dns servers.

The isp's dns servers don't know about your local lan and if asked to resolve a name can't.

This is why you point wkst to your MS dns server which in turn thru the forwarders tab point to the internet dns servers.

None of this explains why everything else works except for the server share. That points to the shares setup.

Have you tried creating a generic share with no restrictions on the server to see if you could see it?

Some things to think about. Best of luck.

Give a person a fish, they eat for a day. Suggest they internet search and they learn a skill for a lifetime.

Ok so what I meant to say was lmhosts file. WINS does work perfectly for this situation, however. I used to work for a newspaper who had ongoing problems with servers not appearing while browsing but they would resolve by name. They had Windows 98, 2000, and XP pc's and all types had the issue. The problem was intermittent, too. I put in a WINS server and pointed everyone to that via the DHCP options on the DHCP server and that resolved that issue. Sure, DNS works great for this too, but if all of your clients are not 2000 or XP, they won't utilize DNS for netbios names, only for internet names. On a a configuration scale, WINS is a lot easier to setup and manage than DNS.

What we need to know is what OS your server is running and what services you already use. Do you have a local DNS or WINS server? Also, have you tried accessing the server via UNC path? Anotherwords, go to start>run and type \\servername\sharename and see if it will come up? If that won't work, just to confirm its not a shares issue, try just \\servername.

However, if you can get to the pc by the UNC path, then the problem is your netbios broadcasting, which I firmly believe is an ongoing bug that MSFT has had for years. Thing is, they've been trying to completely disolve netbios for years so they don't want to fix it. By design, you should be able to stick two pc's on a hub and provided they have statically configured ip's in the same subnet and are using the same workgroup, within minutes they should be able to see each other while browsing the network. This is not always the case. However, you will find that if you use the UNC path, they can talk to each other. That's what we call a "Microsoft feature".

Since the beginning, I've been assuming that you're not seeing the server at all in My Network Places, not just the share. So when you open My Network Places, you dont see the server at all. If you don't see the server at all, that's not a share problem, that's a netbios issue.

One other thing. If you have used a sniffer before, such as Ethereal (freeware), you could install that on your client pc and see if you see the broadcasts coming from the server. You'll see the NETBIOS information in the data section of the broadcast, and it will also include the word "netbios".

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

First, thank you both very much for responding in such a useful way. Now to answer some points raised for the record and anyone's interest:

Actually the Notes server is running on XP Professional I bought an old V4.6 version - many years of supporting Notes and it never causes any trouble.

All machines discussed are running XP with latest fixes. (One of my clients not discussed is running W2000 but symptoms are exactly the same).

I have not intentionally disabled NETBIOS in network bindings. Actually, I don't know how to do this so can't double-check. Can you let me know? Looking at n/w bindings in Advanced for n/w connections, every adaptor has 2 entries: File and Printer Sharing and Client for MS Networks. Each one shows only 1 binding - TCP/IP.

I have never installed NETBEUI. NETBEUI is not routable so assume will make no difference.

I do NOT see the server in Network Places. Once in wireless mode, if I click on the cached server entry, it just times out (for a long time) before coming back with an error.

I am using the DNS server on the Belkin router. I am also using DHCP on the Belkin router. It all seems to be working very nicely. Here the info from ipconfig /all:

Ethernet adapter Wireless Network Connection:

Connection-specific DNS Suffix . : Belkin
Description . . . . . . . . . . . : Belkin Wireless Pre-N Notebook Network Card
Physical Address. . . . . . . . . : 00-11-50-2D-02-10
Dhcp Enabled. . . . . . . . . . . : Yes
Autoconfiguration Enabled . . . . : Yes
IP Address. . . . . . . . . . . . : 192.168.2.4
Subnet Mask . . . . . . . . . . . : 255.255.255.0
Default Gateway . . . . . . . . . : 192.168.2.1
DHCP Server . . . . . . . . . . . : 192.168.2.1
DNS Servers . . . . . . . . . . . : 192.168.2.1
Lease Obtained. . . . . . . . . . : 27 July 2006 14:01:18
Lease Expires . . . . . . . . . . : 19 January 2038 04:14:07

I think that the problem IS a name resolution one but not a DNS one as far as I can see. My standard test for this has always been the following 2 commands which I think proves that DNS resolution works but UNC name resolution does not:

C:\>ping kingpin
Pinging kingpin [192.168.2.3] with 32 bytes of data:

Reply from 192.168.2.3: bytes=32 time=1ms TTL=128
Reply from 192.168.2.3: bytes=32 time=1ms TTL=128

Ping statistics for 192.168.2.3:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 1ms, Average = 0ms

C:\>net view \\kingpin
System error 64 has occurred.
The specified network name is no longer available.

I am happy to configure WINS on my server - I think that's better than static lmhosts file because DHCP might change the ip addresses. I don't think it should be a problem that I have DNS And DHCP servers on the router and WINS on my server should it? I'll try adding the WINS server and get back to you.

Ah, of course, WINS isn't installable on XP so now I'm gutted. V. irritating. I suppose I will either have to upgrade my server to run W2003 Server or else use the LMHOSTS way of working. W2003 is a lot of money just to fix a n/w problem that shouldn't exist so I guess it's LMHOSTS for me...

Adding WINS to your server when DHCP and DNS are provided elsewhere will not be a problem. However in your case I believe your router is just forwarding DNS information from your ISP as I have yet to see a wireless router with a built-in DNS server. So essentially you have no automatic internal name resolution right now.

So to use WINS, make sure to statically assign an ip to the "server" pc, and configure the WINS server address on the other pc's on the network. If you were using a DHCP on your server and not the router, you could have DHCP automatically notify clients of the WINS server. I don't think your router will have that setting, however.

One other thing though before you start implementing WINS. Try using the UNC path with the ip address instead of the name. So go to start>run and type \\serveripaddress\sharename or just \\serveripaddress. So if my server's ip was 192.168.0.1 with a share called "files" I'd type \\192.168.0.1\files on the client pc. This will bypass name resolution and will ensure you can access the server's filestore.

If you can ping the pc by name but you can't access it's shares (even by UNC with the ip address), I'd be sure that you have the windows firewall turned off. You may also leave it on and go to the exception tab and click "file and printer sharing". Make sure the "don't allow exceptions" box on the general tab is not checked.

You did say that "nbtstat -a servername" from the client pc produced a <20> output, right? That should say that the firewall is off on that "server"... but just double check that you can still get that <20> response.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

DOH! Somehow I kept thinking you were using an actual server OS... try the other troubleshooting I mentioned and let me know the output.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

NETBIOS doesn't work either!! I remember trying this before, now - 2 or 3 months ago before I got so frustrated I had to give up and pause for 2 months.

So my lmhosts file now has this line only:

192.168.2.3 kingpin #PRE

You can see from the nbtstat line below that this has taken effect. So now I'll show you what happens - kingpin is the server, pung is another machine on the same wired n/w that shouldn't really be any different. But after connecting my client wirelessly, pung is accessible via NETBIOS, kingpin isn't...

C:\>net view \\pung
Shared resources at \\pung

PUNG

Share name Type Used as Comment

---------------------
C Disk P:
CanonS520 Print Canon S520
Epson Print Epson Stylus COLOR 400 ESC/P 2
SharedDocs Disk
The command completed successfully.

C:\Documents and Settings\jmasters>net view \\kingpin
Shared resources at \\kingpin

MTG Server

Share name Type Used as Comment

---------------------
C Disk K:
SharedDocs Disk
The command completed successfully.

C:\Documents and Settings\jmasters>nbtstat -c

Local Area Connection 5:
Node IpAddress: [192.168.2.6] Scope Id: []

NetBIOS Remote Cache Name Table

Name Type Host Address Life [sec]
------
KINGPIN <03> UNIQUE 192.168.2.3 -1
KINGPIN <00> UNIQUE 192.168.2.3 -1
KINGPIN <20> UNIQUE 192.168.2.3 -1

[[At this point, I disable the wired n/w card and then enable the wireless n/w card on my client]]

C:\Documents and Settings\jmasters>net view \\pung
Shared resources at \\pung

PUNG

Share name Type Used as Comment

---------------------
C Disk P:
CanonS520 Print Canon S520
Epson Print Epson Stylus COLOR 400 ESC/P 2
SharedDocs Disk
The command completed successfully.

C:\Documents and Settings\jmasters>ping kingpin

Pinging kingpin [192.168.2.3] with 32 bytes of data:

Reply from 192.168.2.3: bytes=32 time=1ms TTL=128
Reply from 192.168.2.3: bytes=32 time<1ms TTL=128
Reply from 192.168.2.3: bytes=32 time<1ms TTL=128
Reply from 192.168.2.3: bytes=32 time=1ms TTL=128

Ping statistics for 192.168.2.3:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 0ms, Maximum = 1ms, Average = 0ms

C:\Documents and Settings\jmasters>net view kingpin
[[lengthy timeout]]
System error 64 has occurred.

The specified network name is no longer available.

I am, of course, now completely stuffed as I don't have a solution. Any further advice will be much appreciated.....

thanks,

James.

Thanks for the incredibly prompt feedback.

I've still got the lmhosts going on the client.

Windows Firewall does not run on any of my systems. The Belkin router does my firewall stuff.

But ho! net view using the ip address direct fails on the server (and works fine to my other client). So I think that this means that NETBIOS name resolution is not the problem.

[[So after disabling my wired card and then enabling my wireless card, the following happens]]

C:\>net view \\192.168.2.3
System error 64 has occurred.

The specified network name is no longer available.

C:\>net view \\192.168.2.5
Shared resources at \\192.168.2.5

PUNG

Share name Type Used as Comment

---------------------
C Disk
CanonS520 Print Canon S520
Epson Print Epson Stylus COLOR 400 ESC/P 2
SharedDocs Disk
The command completed successfully.

I can only think that the Belkin router is stopping NETBIOS traffic to the server. By why is it allowing the NETBIOS traffic through to my other client? It makes no sense.

Something v. mysterious is going on - can anyone hazard a theory as to how this can possibly be?

I don't think you are doing the lmhosts files right.

It is not just a matter of editing the lmhosts.sam to lmhosts. You have to go into the nic properties and IMPORT the lmhosts.txt you created at each workstation. This in turn creates the lmhosts file.

Your lmhosts file should include ALL the pcs. You only create one file and use it on ALL machines.

I think you are focusing too much on netbios. Netbios is encapsulated in tcp/ip so it should not be interferred with by the routers switch.

Understand in a peer to peer environment it takes time [up to 15minutes] to build a name list. I suspect you are flying between interfaces and not allowing enough time for this to happen.

The proper setup of lmhosts on all machines should eliminate this issue.

You check that the lmhosts is working correctly by typing nbtstat -c You should see the list as you put it in the lmhosts.txt file

Give a person a fish, they eat for a day. Suggest they internet search and they learn a skill for a lifetime.

http://support.microsoft.com/kb/325487/

might look at this or some of its links

Give a person a fish, they eat for a day. Suggest they internet search and they learn a skill for a lifetime.

You didn't set up any ip filters or any security policies on the "server" box did you?

Try this... remove the line in your LMHOSTS. Do a net view to kingpin from the client while its wired using the ip address. Then disconnect the ethernet and go wireless. Do an "nbtstat -R" on the client. Do a net view to kingpin using the IP ADDRESS. I'm curious if you get the exact same message "the specified network name is no longer available", or not.

Microsoft's article about that error points to all kinds of different possibilities, but none of them could be the issue (all were hardware related) since you can successfully do a net view to another "server" pc over wireless.

Your router isnt filtering that as the wireless and wired switch/hub connections are considered to be in the same physical segment, thus there is no routing taking place and no filtering.

Double check your ip filters on kingpin (TCP/IP Properties, Advanced, IP Filters tab) to ensure that nothing is setup. Make sure you have no other firewall software on kingpin.

If you have a port scanner you might try scanning ports 135 through 139 (of kingpin, by ip address) from the client (while connected via wireless) to see if those show as being open. There are plenty of free port scanners out there. One nice one is called "superscan 4".

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

LOL Wanderer... I looked at the same article earlier, thats the one I was referring to in my message. You're right that netbios is encapsulated inside tcp, but it still uses tcp ports. That doesn't change anything. They got rid of netbeui years back because it wasnt routable and thats the reason for encapsulating netbios in tcp.

This can't be a LMHOSTS or name resolution issue any longer though. Now that he's proven he can't conect to the box via UNC path by ip address (netview \\ipaddress and start>run \\ipaddress), that rules out a name resolution problem.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

? how do you figure \\ipaddress indicates it is not a name resolution issue? That IS one of the signs of no name res.

Only when you can do both \\ipaddress and \\pcname is name resolution not the issue. Name res has to go BOTH ways.

Netbeui is still available for server 2000/2003 [on the cd in the value options folder]so its not "gone".

Before it is said this isn't a name res issue or its not the lmhosts can we at least get the CORRECT lmhosts file on all pcs and see if this resolves the issue???

You can't draw a conclusion from an incompete test.

Give a person a fish, they eat for a day. Suggest they internet search and they learn a skill for a lifetime.

Netbeui isnt "gone" persay, but since it cannot traverse routers, no one is using it.

PC's use ip addressing to speak with each other. They don't use names. Names (whether DNS or NETBIOS) were added for ease of use by humans. When the pc has the ip address, it doesn't reverse it to speak with that box, only if reverse name resolution is requested - which only happens in certain situations.

If you don't beleive me, run a sniffer on a box and connect to another pc via UNC path by ip address. You won't see any name resolution packets - unless you do an "nbtstat -a" or a netview, which asks specifically for that pc's name and share names. In that case, resolution is adhoc and does not involve hosts files or name resolution servers. So if you cant connect to the pc via \\ipaddress, you rule out a names resolution issue.

We can further prove its not a names issue since he can connect to the box via \\ipaddress on the wired connection, he just cant via wireless. There aren't separate names tables for each network interface, there's only one table which is used by all interfaces. Once a connection is made to that box, the information is stored in the netbios cache. The names resolution won't take place the next time a connection request is made, as long as the pc still responds at the same ip address. The pc will try its cache first, hosts & lmhosts files second, and wins/dns servers third.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

name resolution order depends on node type

"PC's use ip addressing to speak with each other"

Incorrect. See here
http://en.wikipedia.org/wiki/Name_resolution
and I quote
"In computer networks, name resolution is used to find a lower level address (such as an IP address) that corresponds to a given higher level address (such as a hostname)"

So what is it going to hurt to actually run the test I suggested? Import the correct lmhosts file to each pc and see what happens.

Yes I agree it seems strange that wired works and wireless doesn't but then I have learned to take what is written and what is actually happening to be two very different things.

This is why we have folks run tests to eliminate the wrong paths so that we may arrive at the correct one.

Give a person a fish, they eat for a day. Suggest they internet search and they learn a skill for a lifetime.

Responding to message 12 by firecodex.

I have checked that there is no IP filtering happening on the router and been through all its settings - there's nothing even slightly suspicious as far as I can see.

I removed the LMHOSTS line and then did your experiment. Here are the results with my comments in []:

C:\>net view \\192.168.2.3
Shared resources at \\192.168.2.3

MTG Server

Share name Type Used as Comment

---------------------
C Disk
SharedDocs Disk
The command completed successfully.

C:\>nbtstat -c

Local Area Connection 5:
Node IpAddress: [192.168.2.6] Scope Id: []

NetBIOS Remote Cache Name Table

Name Type Host Address Life [sec]
------
192.168.2.3 <20> UNIQUE 192.168.2.3 597

[DISABLED WIRED CARD, ENABLED WIRELESS CARD]

C:\>nbtstat -c

Wireless Network Connection:
Node IpAddress: [192.168.2.4] Scope Id: []

NetBIOS Remote Cache Name Table

Name Type Host Address Life [sec]
------
PUNG <20> UNIQUE 192.168.2.5 597
PUNG <00> UNIQUE 192.168.2.5 597
KINGPIN <00> UNIQUE 192.168.2.3 592

C:\>nbtstat -R
Successful purge and preload of the NBT Remote Cache Name Table.

C:\>net view \\192.168.2.3
[LENGTHY TIMEOUT]
System error 64 has occurred.

The specified network name is no longer available.

[END]

I have checked both server and both mentioned clients and there is no IP filtering happening.

I do not have a port scanner and am not too familiar but I'll give anything a go at this point. I'll do this shortly, thanks again for your help.

Re. LMHOSTS, although I agree with the logic of firecodex and can't see how it is a name resolution problem, I also think anything is worth a try, per wanderer's intimation. Who knows, maybe there is name resolution going on at the server end that might be affecting things?

I did not think that one needed to import the lmhosts file to get it to work - certainly it seems to work on XP as soon as the line is added. But I will try as instructed and get back....

I have tried the LMOSTS experiment requested by wanderer. I won't paste any details in - suffice to say that there was no change in the behaviour, regrettably.

I just noticed that the wired connection on the clients had, under the authentication tab to "enable IEEE 802.1x authentication for this network" but the server had this checkbox disabled.

I really thought I'd found the problem this time because I noticed that the wireless connections don't have this tab so I was thinking that perhaps the server was still trying to communicate using the authentication but failing under wireless. So I disabled it on both clients and then switched all 3 machines and the router off. Then switched router on followed by machines but alas, there is no difference in the behaviour.

Yeah the 802.1x authentication can be enabled for a device but it never actually does anything unless the switch you're connecting to tries to authenticate you over 802.1x. 802.1x has to be setup manually and is never enabled by default on any device. It was designed to supplement WEP for wireless because WEP is so easy to hack, but was later ported to wired networks as an access control method. The bank I work for has been discussing implementing it, but I'm trying to convince them to take it a bit further and implement network access control (NAC). Its pricey though. Anyway, enough of that.

So, what do we know? We know its not a name resolution problem, we can assume it's not a filtering or firewall issue (though I would still like to see what the port scan results are), and it seems its not a hardware or drivers issue since you can connect to other servers with the wireless card.

Honestly, at this point I'm at a loss... my only suggestion at this point would be to remove the wireless card, uninstall the drivers, reboot, download the latest ones from the manufacturer and install them, then replace the card. Maybe somewhere there's some corrupt config or driver file that's causing all this. It doesn't seem too likely, but I can't figure anything else. So unless someone else has a suggestion I'd give that a whirl.

Make sure you remove the card and UNINSTALL the drivers - go to the device manager, find the wireless card, right click and do properties, then the driver tab, and select "uninstall". Once its gone, go to add/remove programs and uninstall any programs or drivers that might be associated with it. Linksys and some other brands install their own wireless profile manager apps, so remove those too. Reboot, but before replacing the card, download the newest drivers and install them. Then replace the card. The reason you do it in this order is to prevent any drivers that might not have been properly uninstalled from being used when the card is first discovered.

Good luck...

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

I've tried superscan 4. I don't really understand all this stuff but I attempted to do a scan based on what you said. Here is the result when running in wireless mode.

Live hosts this batch: 1
192.168.2.4
Hostname: [Unknown]
UDP ports (1) 137

______________________________________________

Total live hosts discovered 1
Total open TCP ports 0
Total open UDP ports 1

does that help or should I be doing something different?

You need to scan the target... kingpin.... in your last post kingpin was at 192.168.2.3, and your local pc was 192.168.2.4. So you need to put in 192.168.2.3 as the target. The result you posted shows netbios's main port is open, but I believe you scanned your client, and not the server.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

Well believe it or not the IP address of Kingpin has changed. I've double checked as it sounds odd and the client's IP addresses have not changed but there is no way I could have made a mistake on this. Certainly not anything I did manually - it happened today, I think when I shut all the machines and the router off and then restarted. Apparently DHCP has reallocated the IP address. It makes no difference to anything though - Notes, apache and pinging all still work and net view \\ipaddress works until the switch to the wireless card.

I have just verified (again) that exactly the same thing happens with both client machines. For both, the server becomes inaccessible for net view \\ipaddress following the switch to the wireless card. I've also tried this on my 3rd client which runs W2000 - exactly the same result.

I notice that the IP address for the wired card in my machine is 192.168.2.6 and the wireless card seems to be 192.168.2.7. Could it simply be that the wired IP address is being cached and still used either in the server or the Belkin router? Is there some way to tell this using the scanner tool?

The router doesnt cache ip's. It only stores routes to other subnets. Your pc's are all on the same subnet so that's not the issue.

I'd try reinstalling the drivers using the method I suggested earlier and see if that helps. It can't hurt either way.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

OK, I will try but probably tomorrow now as I'm feeling very frustrated and have spent too much time on this already today.

One more thing for today, I have just changed the server to have a static address of 192.168.2.2 and the DHCP scope now starts from 192.168.2.10. This led me to wonder - if I change both wired and wireless cards in the client to have the same static IP address, I wonder if this would fix it. Obviously, I won't have both cards enabled at the same time. But would it cause a problem?

thanks again for all your advice,

James.

I wouldn't recommend it as the other pc's on the network will cache the mac address of card 1 and associate it with the ip address, then when card 2 comes online with a different mac address you may experience a delay before complete connectivity. Either way I really don't think this will solve the issue since its not a routing issue and you've already proved that the port is open and there's no firewall or ip filters.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

check to see if there is a flash update for the router. consider setting the router back to factory defaults and redoing the configuation. maybe that will kick the wireless into working fully.

Might even want to do a wireless ad-hoc network [wireless card to wireless card] to see if it behaved any differently than when connected via the belkin.

Give a person a fish you feed them for a day.
Ask a person to internet search and they learn a skill for a lifetime.

One odd thing is that there is no workgroup and no master browser registered in the netbios names list (i.e. nbtstat -n)

Netbios codes are:

workgroup = [1E]
master browser = [1D]

type nbtstat -n both from a wireless client and from the "missing" server. What does it say?

Another thing: remove the "#PRE" word from the LMHOSTS entry about Kingpin. That means "use this first before trying other methods of name resolution." Since you seem to be changing Kingpin's IP address (either manually or by DHCP) that is just dangerous.

Check all netbios services are running on the server, e.g. browser, tcp/ip netbios helper.

Reinstall wireless clients and update router firmware as Wanderer suggested.

"If we don't succeed, we run the risk of failure." - BILL CLINTON

He's been doing nbtstat -c's, not nbtstat -a's... so nbtstat -c will only show the cache for that box, it won't show the master browser or workgroup for the target. You have to do "nbtstat -a targetname" for that.

I'd be really surprised if it was the router as he's using the same router for his wired connectivity. The router treats packets coming in the wireless interface the same as it does those coming in on switch ports. The only thing changing here is the interface on the client pc itself. That's why I suggested uninstalling/reinstalling (newly downloaded) drivers for the wifi card.

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+

Re. the nbtstat thing, I've done nbtstat -a against all 4 machines from my client. I then disabled wired card, enable wireless card and did the same again but the results were identical. Note that KONG and IRISNAMESERVER are to do with Lotus Notes. The only point of note is that the Master Browser appears to be my W2000 client. I don't think this matters, though, does it?

C:\>nbtstat -a kingpin

Local Area Connection 5:
Node IpAddress: [192.168.2.11] Scope Id: []

NetBIOS Remote Machine Name Table

Name Type Status
--------------------
KINGPIN <00> UNIQUE Registered
KINGPIN <20> UNIQUE Registered
4WINDSOR <00> GROUP Registered
4WINDSOR <1E> GROUP Registered
KONG <2B> UNIQUE Registered
IRISNAMESERVER <33> GROUP Registered

C:\>nbtstat -a chow

Local Area Connection 5:
Node IpAddress: [192.168.2.11] Scope Id: []

NetBIOS Remote Machine Name Table

Name Type Status
--------------------
CHOW <00> UNIQUE Registered
CHOW <20> UNIQUE Registered
4WINDSOR <00> GROUP Registered
4WINDSOR <1E> GROUP Registered
CHOW <01> UNIQUE Registered

C:\>nbtstat -a pung
Local Area Connection 5:
Node IpAddress: [192.168.2.11] Scope Id: []

NetBIOS Remote Machine Name Table

Name Type Status
--------------------
PUNG <00> UNIQUE Registered
PUNG <20> UNIQUE Registered
4WINDSOR <00> GROUP Registered
4WINDSOR <1E> GROUP Registered

C:\>nbtstat -a trump
Local Area Connection 5:
Node IpAddress: [192.168.2.11] Scope Id: []

NetBIOS Remote Machine Name Table

Name Type Status
--------------------
TRUMP <00> UNIQUE Registered
TRUMP <03> UNIQUE Registered
4WINDSOR <00> GROUP Registered
JMASTERS <03> UNIQUE Registered
TRUMP <20> UNIQUE Registered
4WINDSOR <1E> GROUP Registered
4WINDSOR <1D> UNIQUE Registered
..__MSBROWSE__.<01> GROUP Registered

To eliminate a couple of other points raised, there's no need to worry about the LMHOSTS files now as I've deleted them all. Also, to forestall any queries dynamic IP, KINGPIN the server now has a static address of 192.168.2.2 that will never change.

Finally, yes, I checked the Belkin website for updates some time ago. It lists 2 versions of the router viz:

Wireless Pre-N Router F5D8230-4
with 2 firmware updates:
V1.01.03
V1.01.03 for UK only

Wireless Pre-N Router Version 2000 F5D8230-4_V2000
with only one firmware update viz:
V2.01.02.

My router's base says model:F5D8230-4. However, somewhat ambiguously, it also has a sticker in a different place that says V.2000ea.

The router's web interface reports
Firmware Version 2.01.02
Boot Version 1.00.00
Hardware F5D8230-4 v2

So I'm not 100 pcent sure what model I have but I do seem to have the most recent firmware.

Status report: I have a suggestion from firecodex (21) and from wanderer (reset router and reconfigure) to try. I could try wanderer's wireless to wireless suggestion too but even if it works, I don't think it will help me so leave that til last. I also might try to downgrade the firmware on the router and see if that fixes it. I won't be able to try these for a day or 2 now - will get back then.

I have fixed the problem. I was unkeen to fiddle with the wireless cards - I couldn't see how it could be them because they worked fine across to any machine but the server. The server was the weak link. Most likely was that I'd cocked up the OS or driver settings so first I completely reinstalled XP from scratch. The problem was not gone.

This only left the hardware. So I put a PCI 10 Base T card into the server, disabled the onboard 10Base T port and lo and behold it worked immediately! So the problem is proved irrevocably to be due to the onboard 10 Base T chipset or associated driver on that PC.

For the record, in case anyone else ever has similar trouble, the machine is a Shuttle XPC with mainboard FN41. The 10 Base T chipset on the motherboard is NVidia although no model or number is determinable. The driver which is installed as part of the mainboard chipset installation just says

NVIDIA nForce Networking Controller.

Thanks to all those who advised - it all helped me to hone in on the problem.

Like I said... drivers would seem to be the most unlikely but it was worth a shot to reinstall, wasn't it? I'm glad you found the problem. :-)

---------------------
Firecodex
CCNA/MCSA/MCP/A+/Net+