Can't connect to domain computers until reboot

Question

I have a domain with about 300 Windows 7 and XP machines, with the domain controllers running Server 2003.

Sometimes, I lose the ability to communicate/authenticate with some of the machines until I reboot my computer. This also happens from other computers and regardless of user account and operating system.

In other words, say I'm on ComputerA, and I can't connect to ComputerX. I can go to ComputerB and connect to ComputerX, but ComputerB may not be able to connect to ComputerY. If you reboot ComputerA, then it will be able to connect to ComputerX again.

Here are some messages from various utilities:

• sc: [SC] OpenSCManager FAILED 1722: The RPC server is unavailable.
• mmc (compmgmt.msc): Computer (computer name) cannot be managed. The network path was not found.
• explorer (\\computer): Windows cannot access \\(computer name).
• ping: Packets: Sent = 4, Received = 4, Lost = 0 (0% loss) / Average = 1ms

update:

• netdom verify: The network path was not found. The command failed to complete successfully. (note that this succeeds on the other computers on the network)

Solution

The network works, and you're reporting this as happening on multiple computers. I'd say that you need to look at:

• DNS: Try doing nslookups on the DCs when you run into a failure. Try querying the domain name, and see if you're getting back a valid list of DCs.
• Apps that you run on the PCs that may be too smart for their own good. Especially firewalls, av, etc.
• Network firewall rules. Are you trying to block ports > 1024 between the PCs and DCs? Windows doesn't like port range restrictions.

Comments

[JJ01]

Can you post the results to netdiag

[Excel Kobayashi]

I got " Starting test: Services Invalid service type: RpcSs on (server), current value WIN32_OWN_PROCESS, expected value WIN32_SHARE_PROCESS ......................... (server) failed test Services."

[Excel Kobayashi]

Running the server 2003 version of DCDiag, everything passes

[MikeT]

What kind of events do you see in System event log?

[Excel Kobayashi]

No relevant events in the log

[Excel Kobayashi]

There isn't a network problem. Internet, and other network traffic is fine. Ping still works, the computers can communicate, but cannot authenticate.

[Lemuel]

OK, another question. How many Servers do you have, how many are authentication Servers, and what are the specs of that or those servers. I'm wondering with 300 workstations that maybe you are reaching limits of your servers. Also, how many switches are in your network, is the network still inside of the 432 rule?

[Excel Kobayashi]

We have 3 domain controllers, and about 10 other servers. We don't have any control over the network, but I can tell you it's not very good. Is there a way to tell for sure if the network/server is the problem?

[Lemuel]

3 Domain controllers should be plenty for 300 workstations and 10 other servers. I'm still curious about the switches and whether your network is following the 4-3-2 rule. Granted that's older technology but their are alot of variables and I dont know enough about your network or if cabling as well as switches are up to standards or are they older and slower, which would cause the problems you are having. I'm sorry to be so vauge, I would strongly suggest your supervisors request an audit of your network from an outside source. HTH