This is a Windows 2000 WAN connectivity issue. I have users that connect to our Windows 2000 domain from a remote office that are contantly complaining about connectivity problems. They are on a 10BaseT/Half Duplex UTP Ethernet Segment. They all have Windows 2000 workstations or servers connected to the Windows 2000 domain controllers in the main office.
Here is the network diagram:
W2K Pro/Servers IP ranges 192.168.100.134 to 192.168.100.158 255.255.255.224 Default Gateway 192.168.100.133 (This is the remote office segment.)
| Cisco 4000 router IOS 11.2(16) Ethernet interface 10/half duplex IP 192.168.100.133 255.255.255.224 | WAN link Using IP unnumbered/serial T-1 link | Cisco 4000 router IOS 11.2(16) 10/half duplex IP 192.168.100.132 255.255.255.224 | CheckPoint NG Firewall Ethernet interface 10/half duplex IP 192.168.100.129 255.255.255.224 CheckPoint NG Firewall Ethernet interface 10/half duplex IP 192.168.100.65 255.255.255.224 | Cisco 2600 router IOS 11.3 Ethernet interface 10/half duplex IP 192.168.100.67 255.255.255.224 Cisco 2600 router IOS 11.3 Token_ring interface IP 192.168.0.1 255.255.255.0 | Madge Ring Switch Express | W2K DC 1 Token_ring interface IP 192.168.0.3 255.255.255.0 W2K DC2 Token_ring interface IP 192.168.0.4 255.255.255.0 W2K Appliaction server Token_ring interface IP 192.168.0.5 255.255.255.0 W2K Exchange 2000 Token_ring interface IP 192.168.0.7 255.255.255.0
Each router has static IP routes to the next hop for each host in the remote office. Firewall has static routes for all remote hosts. I can ping by IP and netbios name back and forth from token-ring network to remote office without errors. Same for tracert. I can map drives back and forth, but they disconnect even with autodisconnect registry hack after about 15 minutes or so.
When copying large files from a machine in the remote office, users intermittently get this error: "Error coping file: The specified network name is no longer available." This can happen while a batch file runs on the same machine that loops through pinging by ip the host in the main office. When checking mail from the Exchange server, users intermittently get this error: "A network error is preventing connecting to the exchange server." When running applications on the Windows 2000 application server, connections timeout and apps fail/lockup.
I have applied W2kSPK2 to all machines.
I have enabled lmhost files for the workstations and servers listed above on remote machines.
I have set only one server on remote segment as browse master and maintain server list.
I have run netdiag from Microsoft to show no errors.
I have applied the autodisconnect registry hack.
Sniffers only show reset packet coming back from various servers.
All remote adapters are set to 10/half duplex.
Permitted all protocols to and from remote segment to main office in firewall.
I'm pulling out my hair and getting ready to go postal on this one. I can't tell if the remote network is just too congested with traffic or if there is something else at play here. News groups are worthless on this one. I've seen a bunch of posts, but not one resolution.
Looks like you have a problem related to doing the right thing, which was to upgrade to Service Pack 2. The reset gives me a clue that perhaps the problem is related to a Path MTU problem between your various segments. It turns out that that while this problem is seen mostly with VPNs, the problem you're having is comparable because of the different network architectures you're using. Check out Q301337 for details on how to fix the machines you've upgraded to Service Pack 2.
This was first published in March 2002