Booting a host after deploying image

[joseheitor]

After deploying a Linux image to a new host, the PXE process attempts to reboot, but does not fall through to the GRUB of the newly deployed image. Instead, it continuously recycles, ending with two attempts failing with the following message:

…no configuration methods succeeded (http://ipxe.org/xxxxxxx…)

If I physically switch off the machine and then restart - it loads a FOG menu briefly and then automatically loads the new OS correctly, as expected.

Any suggestions on how to overcome having to physically switch the terminal off and on again?

Thanks
Jose

[george1421]

@joseheitor I can think of 2 possible conditions to cause this.

1. The reboot process leaves the hardware in a strange state where iPXE can’t get an IP address on a warm start.
2. Your network switch and spanning tree (if standard spanning tree) is enabled would cause this. Explanation Default Spanning tree takes 27 seconds to start forwarding network data after a network wink (as if the computer is rebooting). A warm start boots faster than a cold start so by the time iPXE gives up STP has not started forwarding data yet. Where at cold start the computer tests memory and checks hardware that isn’t needed on a warm start. But again this is only a guess as to the reason.

One quick check for spanning tree is to put a dumb (cheap) un managed network switch between the building switch and pxe booting computer, see if that resolves the post imaging reboot. If it does then look into your network switch settings to enable fast-stp, port-fast, or RSTP (whatever your switch vendor calls it).