EFI Pxeboot Failure on 20.04

[Michael Reed]

Recently I noticed this error message for an efi_pxeboot failure.

raceback (most recent call last):
  File "/tmp/nest-_qpddm3y.ddffe1464fb1472263c562eaa13feeb6fd7958c5aef597cea6e3f81f7455786c/efi-pxeboot", line 150, in <module>
    sys.exit(main())
  File "/tmp/nest-_qpddm3y.ddffe1464fb1472263c562eaa13feeb6fd7958c5aef597cea6e3f81f7455786c/efi-pxeboot", line 145, in main
    retval = is_pxe_booted(boot_entries, boot_order, boot_current)
  File "/tmp/nest-_qpddm3y.ddffe1464fb1472263c562eaa13feeb6fd7958c5aef597cea6e3f81f7455786c/efi-pxeboot", line 84, in is_pxe_booted
    desc = boot_entries[boot_current]
KeyError: ''

[Michael Reed (mreed8855)]

I will request the output from the following command as I to not have access to the system:

sudo efibootmgr -v"

[Rod Smith (rodsmith)]

This has been spotted more recently on a Lenovo component submission:

* https://certification.canonical.com/certificates/2111-12794/
* https://certification.canonical.com/hardware/201904-26996/submission/236757/test/67646/result/21400845/

Traceback (most recent call last):
  File "/tmp/nest-80086b4h.e641870c9d3cd25e04d96a73a598df2a1012f8b6825e9e105512234744aebb13/efi-pxeboot.py", line 150, in <module>
    sys.exit(main())
  File "/tmp/nest-80086b4h.e641870c9d3cd25e04d96a73a598df2a1012f8b6825e9e105512234744aebb13/efi-pxeboot.py", line 145, in main
    retval = is_pxe_booted(boot_entries, boot_order, boot_current)
  File "/tmp/nest-80086b4h.e641870c9d3cd25e04d96a73a598df2a1012f8b6825e9e105512234744aebb13/efi-pxeboot.py", line 84, in is_pxe_booted
    desc = boot_entries[boot_current]
KeyError: ''

At a first pass of examining the code in efi-pxeboot.py, it looks like the BootCurrent EFI variable might be missing or damaged in some way that's causing Python to error out.

[Jeff Lane  (bladernr)]

Got this dump of efibootmgr -v from the machine failed in comment 2 (though this was from a 21.10 install as they had already moved on from this)

[Rod Smith (rodsmith)]

I've just checked with the efibootmgr.txt output you've posted, Jeff, and the current script handles it fine. (I had to hack to script to take a text file as input, though.) That output fails the test, but legitimately so, and the script does NOT crash with a Python traceback, as the bug describes.

The efibootmgr.txt output DOES include a BootCurrent variable, the lack of which is what I believed was causing the crash. Thus, my suspicion is that BootCurrent was originally missing but has appeared on this machine for whatever reason; however, it's conceivable that something else was causing the crash. If so, then either the already-merged bug-fix branch has resolved the problem or the efibootmgr.txt output that Jeff has attached does not show the cause (presumably because something has changed in the output between July and today).

[Jeff Lane  (bladernr)]

Thanks Rod, I'm gonna close this. It hasn't been seen in quite a while, so as we discussed I suspect either your fix merged in and/or changes in the OS have resolved it. If it appears again we can open a new bug and deal with it there.