I spend 3 months as a Problem Analyst. Basically the organization had a lot of problems (as defined by ITIL) and one of their annoying issues for Deskside Support / Administrators was imaging PCs /Servers, basically they would not start image reliably…
Problem:
You add a new device into SCCM via the Computer Association (say – Computername and MAC address)
You rush off to the device and turn it on and PXE boot it
Absolutely nothing happens (PXE boot fails with an abortpxe.exe)
You rush back to the SCCM Console and check the collection, yep it exists, so you check the “All systems” collection, it may exist, or it may not, so you do and “update collection membership”. Yep its in all systems, so it must be good!
You rush back to the PC and PXE boot it and…you get an abortpxe.exe error again…
What’s going wrong!!!
Reasons:
- The users were rushing things in SCCM. SCCM took it’s time to make the new association available to the PXE server
- The PXE server was not checking back with the SCCM server as the PXE cache time was longer than the time between boots.
Solutions:
- Before you boot the device , make sure your device shows in “All Systems”
- There is a “feature” with different service pack levels of SCCM 2007 around the behavior of the PXE server make sure you know what your PXE cache expiry is set to . So if you boot the device, don’t try to boot it again until the PXE cache has expired…..really…. don’t …it won’t make a blind bit of difference.
Notes:
- Setting the value in CacheExpire is only supported in ConfigMgr 2007 SP2. (something like 2 minutes is fine). It is a dynamic change, so no service or server restart required, in my experience
- Whilst you may change the CacheExpire… changing the network adapters may well set the CacheExpire to the default. I have seen this behaviour when the physical blade was moved to a different chassis (well this is what I suspect, no ability to confirm)