Re: Re GRUB timeout not working after power failure

[John Little]

On 5 December 2017 at 10:00, Benjamin Moody <address@hidden> wrote:
> It occurred to me, after writing that first mail, that for some reason
> there might not be a "default" entry in this situation.  I had looked
> at the 'grubenv' file and seen that (at least while the system was
> running) there was always a 'saved_entry' defined.  So I didn't
> initially consider the possibility that 'default' might be undefined
> or point to a nonexistent entry.

This is why I say it's a CentOS issue; I imagine that CentOS has a
mechanism as part of shutdown that writes to /boot/grub/grubenv, to
give the behaviour you describe.  I'm on a debian-derived distro, and
it has an rc script that unsets a grub variable on startup, to signal
that a system has successfully booted:

    grub-editenv /boot/grub/grubenv unset recordfail

> I can... sort of... understand the motivation behind making 'saved'
> the default, but for servers it seems like a bad idea in any case.

I eventually decided that the debian thing was there to stop a system
continually rebooting, when a boot fails.  Perhaps the CentOS
mechanism has a similar motivation.  I can imagine servers for which
that would be bad, and lead to hardware failures.

Hence, searching CentOS documentation or wikis might help.  Also, I
suggest you cd /etc and search for files with grub in their name, you
might find out what's going on. In any case, if you know the state of
grubenv you want, you could add a start-up script that does a
grub-editenv, or whatever CentOS supports.

Regards, John Little