Troubleshooting NET3226 Errors (99255)

MORE INFORMATION

While knowing that the problem stems from the primary domain controller (PDC) narrows the problem somewhat, it still does not adequately cover all possibilities. The first thing to do is translate the hexadecimal "nnnn" of the number 0000:nnnn associated with the message. By way of example, here is an explanation of a value for 'nnnn' taken from a recent incident: C708.

First reverse the order of the two pairs of hexadecimal numbers:

C7-08 becomes 08C7

Next, translate this into decimal by multiplying the first position by 4096, the second position by 256, the third position by 16 and adding all of these together along with the last position:

0 * 4096 =    0
8 *  256 = 2048
C *   16 =  192 (since C in hex is 12)
7        =    7
           ____
           2247
				

Now use this number in net helpmsg 2247 to produce: This clarifies the problem and helps you determine what to do. If the number computed is far out of range (valid values are approximately 1 to 5500) or seems inapplicable to the situation, go back and make sure you transposed the pairs and performed the conversion correctly. For instance, if C708 is converted without being transposed, the value is 50952.

Not all net 3226 messages are a result of the above problem. These causes were extracted from incident reports to Product Support Services in the last year.

• The day and time of the primary domain controller was more than 10 minutes different than that of the backup domain controller or member server.
• 'Numbigbuf' was set too low, UAS replication uses this resource. The default should be adequate. Unfortunately, a documentation error suggests setting the value to 1 in LANMAN.INI if only HPFS is used. Remove this entry if you find it set this way.
• The primary domain controller was paused.
• Little or no free hard disk space was available.
• 'Sessions' had been set too low.
• Tuning problems were causing time-outs.
• A network card or repeater was bad.

Diagnostic and corrective actions may include:

• Make sure the PDC has accounts for all BDCs and members included in its Servers group.
• Audit logons to see if failure to replicate is a result of the BDC or member not being able to logon to the PDC.
• Run the CHKUAS program included in the Resource Toolkit to check for UAS errors.
• Run the SYNCSTAT program (available from Microsoft Product Support Services) to get further information about synchronization status.
• Synchronize the date/time of all servers.

At the problem server:

• Demote the problem server to stand lone and promote it again.

For more serious backup or member UAS corruption:

1. Decompress NET.AC$ in the ACCOUNTS subdirectory on the OS/2 Server diskette by doing DECOMP A:\ACCOUNTS\NET.AC$ at the server.
2. Demote the problem server to standalone.
3. Shut down the problem server.
4. Boot it with the HPFS Recovery disk.
5. Rename NET.ACC, then rename NET.AC$ to NET.ACC--if it is in the <lanman>\ACCOUNTS directory.
6. Reboot the server.
7. Promote it to backup domain controller or member as appropriate.

For version 2.1a, a patched NETLOGON.EXE (available in CSD01.044) corrects one cause of the net 3226 error.