PXE boot from USB NIC

Thought I would post what the solution turned out to be. WDS PXE service had a "wait time before responding" that someone had set to 30 seconds. I noticed that vmware ESX vms only polls PXE for maybe 10 seconds. ... they just weren't waiting long enough.

I lowered the "wait time before responding" setting, and all is well.

We faced the same problem and found the following to be excellent for explanations:

• http://gparted-forum.surf4.info/viewtopic.php?id=16400
• http://sourceforge.net/projects/clonezilla/forums/forum/663168/topic/4511817

Our configuration:

• Debian/kFreeBSD (sid)
• tftp-hpa (5.2-4)
• Clonezilla Live Image (current stable 1.2.12-67)

The cause of the problem, like explained in the second link, it's related to the value of the -b parameter (blocksize option) of the tftp command. The tftp command is used to retrieve the squasfs file system from the server, and it's invoked (in this case with the wrong parameter -b value) by the live script in the initrd.

Our workaround is to modify the initrd editing the inside live script (/scripts/live) and adjust the problematic parameter value no bigger than 9212 or delete it.

Replacing:

tftp -g -b 10240

with

tftp -g

fixes our problem.

To some good explanation about modifying an initrd see:

• http://blog.mybox.ro/2010/05/12/how-to-change-files-in-a-initrd-image/