I ended up opening a support ticket for this referencing this thread. I even created a new OSD manager on a separate server and the error messages persist.
Can someone tell me how BMC does TFTP? It looks like it previous versions there was a separate TFTP application you had to install but it looks like that is no longer the case and the application is supposed to have a built in one?
I have options 66 and 67 setup on my target subnet pointing to the local network listener/image repo and boox64.efi. I have also setup WireShark on the local network listener/image repo and performed packet captures between the listener and the target pxe workstation which are both on the same subnet. This is essentially what happens:
Source Destination Protocol Info
Boot listener Target Wks TFTP Read Request, File bootx64.efi Transfer type: ocetet, tsize=0, blksize=1468
Boot listener Target Wks TFTP Option Acknowledgement, blksize=1468, tsize=99128
Boot listener Target Wks TFTP Error code, Code: Option negotiation failed, Message: User aborted the transfer
Boot listener Target Wks TFTP Read Request, File bootx64.efi Transfer type: ocetet, blksize=1468
Boot listener Target Wks TFTP Option Acknowledgement, blksize=1468
Boot listener Target Wks TFTP Acknowledgement, Block: 0
Boot listener Target Wks IPV4 Fragmented IP protocol (proto=UDP 17, off=0, ID=4260)
Boot listener Target Wks TFTP Data Packet, Block: 1
Boot listener Target Wks TFTP Acknowledgement, Block: 1
Then it repeats the last three lines
Deploying a sysprep image is unsuccessful as it continually tries to use multicast even though I told the job not to use it. Image captures on the same workstation is successful....
I have been working with Steve Gibbs and customer support on this. Steve asked me to setup one of my local workstations as an OSD manager that is on the same subnet as the target pxe box. Workstations are on the same switch and same subnet. The issue still persists. I took it further and setup two OSD manangers with one running ADK 1903 and the other 1607, they both have the same unicast error messages but where they differ is the load time of the BMC PXE menu to get into Win PE. The OSD with 1903 takes 1-3 minutes from the time the BMC PXE menu shows up to boot into WinPE while the OSD with 1607 takes 15-20 minutes and these were simply Test Projects only. They perform their diskpart operations then the unicast error messages appear.
Something else I noticed:
When using ADK 1607 on an OSD manager more often than not any project that gets generated will fail with the error “ failed to get windows\system32\osdhelper in the wim. Sometimes I can fix this error by having the OSD configuration do a full reinitialization. Sometimes by moving the PXETFTP folder or reinstalling Win ADK 1067 but most of the time it keeps failing. I notice that this file gets inserted into the WIM file by looking at the init86 and amd64 logs within the OSD log folder but it doesn’t appear to work.
I am going to test the other versions of ADK such as 1703, 1709, 1803, and 1809.
I've also opened a ticket with support.
Hey Raul Caballero, are you able to successfully deploy a sysprepped image w/o the unicast transfer errors?
Patrick Berrigan, I'm using the 1703 ADK if that helps,
So far, the deployment will roll through the unicast errors, but without the unattend files the computer is unable to boot.
I'm going to try capturing another image next.
Yes I have, so after removing the OSD folder and roles and copying one from a 12.9 regular client then reattaching the roles and building the projects, target lists, and drivers I was able to get the Master/OSD manager to work properly. I have not yet gotten a response as to why my Relay continues to have the issue. I currently have a support ticket with them which they haven’t given me any insight in but I am SURE 1903 ADK works as it worked before the upgrade, and now works on my Manager connected computers. Don’t want to point fingers or anything but based on the resolution I had to come to and the fact that 1903 indeed does work I am under the impression they do not fully understand how their system is affected by the ADK. I forgot to mention, I uninstalled the ADK as well and reinstalled it along with the PE package because I know the install wim that BCM uses is built from that. My hypothesis is that the install wim for building the projects got corrupted and therefore by forcing a rebuild by reinstalling the PE environment as well as forcing it (purging it) to rebuild the install wim it was able to resolve the issue. I’m not sure how the relay comes into play or if it uses a cached wim but my best guess is there is still some lingering error. Can anyone confirm that a ground up setup works properly?
Would you all please post your final thoughts and solutions that were used to resolve this issue? I am getting directed emails asking me if I know what the "fix" is as more folks are moving toward v12.9.x.
Patrick, I know you and had a call about another issue and I asked about this issue and you indicated you got it working but due to my many conversations with lots of folks I cannot recall your answer.
Raul, If you have indeed resolved this issue can you please mark your final thoughts and resolution and the correct answer so folks can use your experience as a self service solution so others can benefit from your thread?
Richard, same goes for you... Hopefully you all will have the same response as to the resolution of this issue but it would benefit all to have a conclusion on this discussion post.
I want to thank you all for your participation in this community in efforts to not only benefit yourselves but others facing some of the same issues.
If anyone from the BMC Client Management support team can also provide final thoughts on this post - please share your notes too!
My issue was resolved by loading the correct WinPE driver.
My issue with the is ongoing, support has me using a modified osdeployment.dll. it seems slightly more intermittent.
The only pattern I can see is that "we won't use multicast transfers" & everything goes through."we may use multicast transfers" and I start to see failures.