Post SP6a Fixes

Christopher Rice

This article describes issues not covered on the latest service pack for NT

I found the following information that I felt compelled to share with you all, as I have heard of no new service pack coming for Windows NT. I am taking a portion of an article from the Windows 2000 magazine:


   I found the following information that I felt compelled to share with you all, as I have heard of no new service pack coming for Windows NT. 

   I am taking a portion of an article from the Windows 2000 magazine:


    QLogic Fibre Channel driver crash. Under heavy system loads, a multiprocessor computer using a QLogic Fibre Channel disk array controller might experience Stop 0x0a blue-screen crashes from scsiport.sys and ql2100.sys drivers. According to Microsoft article Q264247, the stop occurs when the scsiport.sys returns a virtual address that belongs to the QLogic PhysicalCommonBuffer. The article indicates that a stressed system will crash once or twice a week. To eliminate the problem, contact Microsoft Support for a post-SP6a version of scsiport.sys released on May 15.

RAS clients can't reconnect. Many readers have emailed me about RAS clients that can't reconnect after they have established a working connection-and I finally have one possible explanation for the reconnect problem. Microsoft article Q256507 states that if-after a successful RAS logon-you disconnect and then immediately reconnect the client, you might see the error message, "A domain controller for your domain could not be contacted. You have been logged on using cached account information. Changes to your profile since you last logged on may not be available." The problem occurs when the RAS server assigns a new TCP/IP address to the client when the client reconnects. The authenticating domain controller doesn't update its local NetBIOS name cache with the RAS client's new TCP/IP address, so the domain controller can't authenticate the RAS client. There's a good chance you can fix the RAS client reconnect problem if you call Microsoft Support and ask for the netbt.sys update released on March 2.

Nbtstat problems on RAS servers. When your RAS server has a large number of network interfaces (i.e., 30 or more dial-out lines), you might experience two NetBT-based problems. (NetBT is the driver that manages NetBIOS over TCP/IP connections to and from the local computer.) First, when you run nbtstat.exe from a command prompt, the command won't return any results (except another command prompt). Second, when you run Performance Monitor, you might see Event ID 3101 in the Application Event Log (AEL) with the text "Unable to read IO control information from NBT device." Microsoft article Q259241 indicates that NetBT misbehaves when it reads bindings information from the Registry into an internal buffer and the Registry information exceeds the size of the buffer. The buffer overflow condition is more likely to occur on systems with 25 or more network interfaces bound to NetBT.

The article offers two workarounds for the problem. You can unbind NetBT from all the interfaces that don't require NetBIOS, or you can reduce the number of dial-out lines until you no longer experience problems. To correct the problem permanently, call Microsoft Support and request the bug fixes (a new nbtstat.exe and a new perfctrs.dll) released May 24. The Microsoft article doesn't categorize this bug fix as post-SP6a, so I assume the problem affects earlier versions of NT as well.

HP print driver crashes. Are you seeing blue screens with a stop code of 0x0000000a or 0x00000050 that attribute the problem to HP print drivers? Microsoft article Q262492 indicates that outdated drivers for HP printer models 4500, 8000, and 8001 can cause these system crashes. The article documents a diagnostic procedure you can use to make sure that outdated drivers are causing the problem. To perform the procedure, you must run the pstat.exe tool from the Windows NT Server 4.0 Resource Kit. Pstat.exe displays a list of installed DLLs. If your system has outdated HP drivers that cause such crashes, the list will include a phantom ntdll entry.



*****The entire article can be found at the Windows 2000 magazine site under ArticleID=8969.

This article was originally published on Jun 8, 2000
Page 1 of 1

Thanks for your registration, follow us on our social networks to keep up-to-date