Windows networking – LAN broken if no Internet

Topic

New Reply

small network – 8 or 9 windows XP pc’s – peer to peer with unix server using samba to share some shares from unix to windows (just directories as network drives).

all works fine unless the internet goes down (which it does frequently there!).

if the internet is down – they can not browse the network (neither the windows PC’s nor the unix server) or access the network drive shares for purpose of file sharing.

yet they CAN print to network printers, they CAN login to unix via their telnet client, they CAN http to unix server and their router.
they CAN ping themselves as well as any other PC/server/printer on the network.

the PC’s have a mix of trend-micro and mcafee antivirus/firewalls, i tried shutting the firewalls to no avail, and because they are different brands i think i can eliminate the A/V.

Now – this only occurs if the PC is booted with internet off. If internet is down, PC booting process takes a LONG time (8 mins?) to boot.
does not give any errors and nothing logged in event viewer.

But – if internet already running when PC is booted for the remainder of time until PC is rebooted – if internet goes down, all still works fine.

Something is happening when windows looks around at boot time that is choking it. Even if internet restored, it can not recover without a reboot (tried disable/enable adaptor, did not recover, see nothing weird in ipconfig /all).

Nameservice probably not the culprint as can’t even get away with \some.internal.ip.address

I can readily duplicate by pulling the DSL cable before it hits the DSL modem. I can’t make it happen here in my office on my network with my PC’s (e.g. if i reboot with internet disconnected).

Currently, cabling is:
DSL/DSL Modem -> Linksys Router -> Linksys 12 port hub.

This is a big problem for them as the store nothing on local PC’s, all is stored on shared network drives.

Reply | Quote

Replies

Sounds like a router problem to me, which could be causing both the frequent internet disconnections and the local network routing problem. Have you got another router you can try? And what kind of hub do you have, just a dumb hub or a switch? If this doesn’t work post the exact models of router and hub and the network topography.

Reply | Quote

Sounds like a router problem to me, which could be causing both the frequent internet disconnections and the local network routing problem. Have you got another router you can try? And what kind of hub do you have, just a dumb hub or a switch? If this doesn’t work post the exact models of router and hub and the network topography.

Its not the router that is causing the internet to fail, it is the ISP.

When the internet is down, i don’t even know why the router would be involved in a local LAN address issue anyway. Power cycling it had no effect. I am not on-site right now, but i’d guess it is a linksys BEFSR41 or BEFSR81. Hub is an intelligent switch – linksys EZXS16W from my notes.

Reply | Quote

I know it’s dangerous to assume, but I assume you are using DHCP and the router is assigning the IP addresses to your PC’s. With PC’s taking a long time to boot it sounds as if it’s trying to obtain a DHCP IP address and eventually failing … which is most likely a router (or very slight possibility of switch) problem.

From the description I am also dangerously assuming you do not have “Enable NetBIOS over TCP/IP” checked off in your advanced network settings. If that’s the case then the switch will not route the NETBIOS traffic and it’s up to the router to do so, which can put the problem back on the router.

Reply | Quote

No, all IP addresses are static, there is no DHCP. All PC’s _are_ set for netbios over TCP/IP enabled.

Reply | Quote

No, all IP addresses are static, there is no DHCP. All PC’s _are_ set for netbios over TCP/IP enabled.

OK, I am making another assumption that all the PC’s are plugged into the switch and not using port 16. And here’s the tech support person in me, so bear with me … what are the IP addresses the PC’s are set up for, their netmasks, gateway addresses, and DNS servers?

Reply | Quote

I doubt they are on port 16, but not 100% sure. if they were incorrectly using some uplink port it would never work, not this wierdness.

PC addresses are 192.168.1.20 to 40, network printers are 192.168.1.90 to 99, unix/samba server is 192.168.1.140, gateway is 192.168.1.50 – netmasks are all 255.255.255.0

for DNS, i believe each PC has two 64.something addresses from their ISP, plus a 3rd from OpenDNS – probably 208.67.220.220 – i checked these when checking network settings, but did not write them down and can not access remotely to check, but you get the gist of it.

Reply | Quote

192.168.1.50 is a strange number for a gateway address, I would have thought it would be 192.168.1.1, but if it works, it works.

The slow boot when the Internet connection is down is probably due to some service trying to do a DNS lookup and can’t reach a DNS server. I would try putting the router’s IP address (the same as the gateway address) in as one of the DNS servers. How do you connect to the local shares, via IP address or DNS/WINS name? If not by IP address, try using that instead.

Reply | Quote

> 192.168.1.50 is a strange number for a gateway address, I would have thought it would be 192.168.1.1, but if it works, it works.

i always change from 192.168.1.1 to something else with any network appliance – too many things come preset at that number and you just get headaches when plugging them into the network when they conflict.

> The slow boot when the Internet connection is down is probably due to some service trying to do a DNS lookup and can’t reach a DNS server. I would try putting the router’s IP address (the same as the > gateway address) in as one of the DNS servers. How do you connect to the local shares, via IP address or DNS/WINS name? If not by IP address, try using that instead.

i did try by IP – just running \ip.address.of.pc hangs forever when network is in this condition (but works if internet connection was up when PC was booted)

Reply | Quote

Is the DSL modem also a router or is it strictly a modem? My only thought was that if it is also a router and DHCP wasn’t disabled or assigned solely as the Linksys router gateway, there might be a mixture of incidences when the Internet fails that cause the modem’s DHCP to become predominate, leaving the network high and dry.

If its just a modem, never mind.

Reply | Quote

the DSL modem is also a router, but ISP controls it, doesn’t let you in – because of that we had them turn off NAT and DHCP, and we use the LAN address of it as a WAN address for our own router where we can do NAT, port forwarding, etc. the ISP has had techs work on it so its possible they enabled DHCP again, but even if they did – since we are using static IP and their DHCP would be on a different subnet, i don’t see how it would affect the LAN as it is setup, since all the static settings (IP, DNS) protect it from that

Reply | Quote

It could be that your NETBIOS is set to the wrong node type(s) and is never able to resolve the peer addresses properly.

Here is an explanation of the NETBIOS nodes:

http://www.tech-faq.com/netbios-node-type.shtml

It looks like you would want your computers to be set as Hybrid (H) or Peer to Peer (P) nodes based on the above link.

Reply | Quote

Not being able to connect to a PC by specifying the IP address (\ipaddr) sounds like a bad default gateway setting.

Do you have the default gateway set to your own router or the ISP router? It should be your own, with routes set in your router to send internet traffic via the ISP router.

cheers, Paul

Reply | Quote

the gateway address is my own (internal) router.

Reply | Quote

the gateway address is my own (internal) router, and besides – not only should internal ip’s not have to go through the router, but it would affect all protocols if that were the problem, and things like ping, http, telnet, etc, which do work would then fail too.

Reply | Quote

Have you tried a different router and/or the standard #.#.#.1 gateway address with the same results? Its hard to troubleshoot well since you can’t also get into the modem. Nothing on the network has a static I.P with a lower number than the gateway does? And the router has had the I.P ranges adjusted or limited accordingly? Seems to keep coming back to the router having a spaz when the Internet goes down ultimately, either from some setting or malfunction.

Reply | Quote

i have now setup the linux box as a WINS server, and configured all PC’s networking to look for it as a WINS server, thus forcing node into hybrid mode, checked that all were enable netbios over tcp/ip. still no joy.

tried resetting router while in this condition – did not make a difference (and if i am looking at internal IP’s, i’m still not sure why it’d even be hitting the router). i can’t swap router remotely, but i might be able to try one in 2 or 3 weeks if i go out there.

PC addresses are 192.168.1.20 to 40, network printers are 192.168.1.90 to 99, unix/samba server is 192.168.1.140, gateway is 192.168.1.50 – netmasks are all 255.255.255.0 – don’t see why PC’s being higher/lower numbers than gateway address should make a difference.

this is a tough one

Reply | Quote

Have you tried setting the gateway (your router) to handle the DNS and point all the connecting PC’s to 192.168.1.50 for DNS?

(My router is xxx.xxx.xxx.254 = default setting and everything else is lower, can’t see that as an issue for you.)

Reply | Quote

i don’t know how DNS comes into play if i cant click \192.168.1.140

Reply | Quote

Certainly is odd. Maybe something on network settings is set to point to an external address and when it’s not available the network is kaput?
You could try Wireshark or similar network sniffer to look at the packets from one of the dodgy machines to see what it’s doing when trying to share files?

cheers, Paul

Reply | Quote

i’ll be out there in about 2 weeks or so, will add ethereal or wireshark to whatever i try next to debug a bit more

Reply | Quote

wireshark didn’t show anything good, also tried 2 new routers which did not help.

Reply | Quote

just to close this out – turned out to be samba problem – shares on windows boxes in the network did come up eventually, but after a LONG, LONG, LONG delay.
The fix was to add local DNS/nameserver (done on unix box), after which samba stopped caring that outside DNS did not work – apparently either recent Samba, Bind, or both, require DNS to answer in order to work.

Reply | Quote

Windows does not handle delayed network response, it makes the whole machine crawl – hope W7 is better.
You had the workstations connected to samba shares and Windows shares?
Did it make a difference which mapping came first?

cheers, Paul

Reply | Quote

> Windows does not handle delayed network response, it makes the whole machine crawl – hope W7 is better.
thats for sure. haven’t tried W7 in this mix yet

> You had the workstations connected to samba shares and Windows shares?
yes, both

> Did it make a difference which mapping came first?
no.

Reply | Quote