Shulou(Shulou.com)06/02 Report--

Recently, I did a SCCM2012R2 project, which made me physically and mentally exhausted. I encountered all kinds of pits. Today, I would like to share with you some pits and related solutions encountered in this project.

The first phase: PXE startup failure troubleshooting

Problem description:

=

When PXE is done across network segments, there will be a blue screen, and the PXE is normal under the same network segment.

Troubleshooting:

=

From the error point of view, WDS no response, we suspect that the function of WDS is abnormal. We try the following steps to re-enable WDS, and the problem remains:

-

Log in to the WDS server and stop the Windows deployment services server service

Find the RemoteInstall\ Mgmt path and cut all the files under that path to a new folder for backup

Restart the Windows deployment services server service, and the files under the Mgmt path will be regenerated

Then we grabbed the network packet on the DHCP server and SCCM DP, and we can see that when reading the BCD file, the file could not be found, and the failure to find the BCD file directly led to the initial blue screen error (boot configuration data is missing or contains errors).

-

The sequence of normal PXE boot execution is: Wdsnbp.com--Pxeboot.com-Bootmgr.exe-BCD file-Boot.wim

You can see that the Wdsnbp.com file was read successfully in the network packet of the faulty machine.

Pxeboot.com file read successfully

Bootmgr.exe file read launch successfully

Then when you read the bcd file, it shows that the file cannot be found.

3. After communication, we learned that the customer configured DHCP option 66JI 67, because Microsoft does not support this situation, so we recommend that the customer configure IP helper instead. After configuring IP helper, we encountered the following new error.

4. We have done a packet capture analysis of the new error report, and found that the following analysis

5. After we get the customer's network topology, we come up with the following packet capture scheme:

-

Enable port mirroring on the interface where the client accesses the S2126G-6 switch and enable netmon on the corresponding machine

Enable port mirroring on the interface between the WDS server and the core S7806A switch and turn on netmon on the corresponding machine

Enable port mirroring and turn on netmon on the G1Acer 42 G1Acer 45 of core S7806A and the G0Acer 24 interface of converging S5750A

Open netmon on the WDS and DHCP servers.

Note: we did not grab the packet together with the port on the access layer switch and the G0swap 17 of S5750A because we considered that it was not very likely that the problem would be in this section.

6. After waiting for about 4 months, you grabbed the package according to our plan and submitted the log. The problem has changed again in the latest network package, which is different from the previous two times.

-

After the client gets the wdsnbp.com, the file returned by the DHCP request sent again is not a normal Pxeboot.com, but an abortpxe.com, which directly terminates the PXE boot behavior.

7. After that, we did a remote together, and we found the following checks:

When deploying "Guangdong Lenovo MT4500T_Win7_X86_SP1" to "all unknown computers", it was found that the operating system image no longer existed when this task sequence was broken. We fixed this task sequence and re-added the system scene of Windows 7SP1. After that, we successfully deployed this task sequence to the "all unknown computers" collection.

After that, we found two physics notebooks to do the test. When booting through the network, the current error message is completely different from the previous error, there is no blue screen (BCD Error) or PXE-M0F error. The current error message is:

Because the client needs to request policy from SCCM MP, we check the MP-related logs and find that: MP has been unable to connect to the site database for search, and the status of MP is abnormal; we also found a large number of persistent DB unable to connect errors and Internal 500errors in mpcontrol.log; we connected to port 1433 of the site database on the SCCM DP server and found that we could not connect, which is the problem of abnormal MP operation.

8. According to the error we found in the remote conversation, you opened ports 445 and 1433. After testing again, it is found that the client is still requesting policy from SCCM MP. We recommend that you follow the steps below to PXE re-enable, and test the PXE boot successfully after completion.

--

SCCM console-Management-Distribution points

Find the 20GD distribution point, open the base properties, cancel the PXE-related settings, and confirm the application.

Wait 15-20 minutes and restart the server.

Delete the original RemoteInstall directory (make sure there is no such directory under each partition, if so, please delete or rename it)

On the SCCM console, locate the 20GD distribution point, and then enable PXE. The system will reinstall PXE (WDS)

Wait 5-10 minutes, confirm that the WDS service is installed (confirm the system service deployment service and the RemoteInstall directory in the data partition), and test the PXE boot after installation.

For the troubleshooting of these two parts, according to my analysis and experience, the reasons for PXE startup failure are as follows:

A. Boot image exception on the PXE/WDS node-We specifically deleted the old folder in the subsequent PXE reinstallation to ensure that the reinstalled PXE is not affected by the previous old files.

b. Deployment exception-there is no sequence of tasks previously deployed to all unknown computers, which indirectly causes the corresponding task not to be found when PXE starts.

c. Abnormal port shutdown-as a result, during the PXE startup phase, the site server could not complete the query normally and the task sequence could not be detected.

The second stage: task sequence is stuck.

9. After solving the PXE startup problem, the system has successfully started from PXE to WinPE, but did not see the task sequence as expected in the WinPE phase, and a blank interface has been displayed as follows:

Since the log of the WinPE phase could not be obtained from the field machine: smsts.log, we tried to test this problem through the virtual machine. In the process, we adjusted the setting on DP so that it can be debugged during the WinPE phase (call up the command console by F8)

10. When passing the virtual machine test, we received the following error:

This error message is caused by a damaged or faulty WinPE image. We re-establish the Boot image and adjust the Boot image setting in the task sequence to ensure that the error disappears after the corresponding task sequence matches the operating system. However, the WinPE phase is still the same as the figure above, showing no task sequence.

11. In the smsts.log logs collected by the virtual machine, we found the following information. It shows that during the WinPE phase, this client machine cannot properly connect to the corresponding MP and DP servers.

This problem was quickly resolved after my network colleagues adjusted the DNS settings. You can also see the task sequence normally.

At this point, the system can be deployed normally, and the test results on the virtual machine are also normal.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.