Bug#783081: installation-reports: Jessie daily installer installs kernel in flash without any further confirmation (QNAP TS-212)
On Thu, 2015-12-10 at 14:25 -0800, Martin Michlmayr wrote:
> * Bernhard Übelacker <bernhardu@vr-web.de> [2015-04-21 21:00]:
> > But I was not aware that the Installer would overwrite the kernel
> in
> > the internal flash without further asking. (Most probably because
> > the internal flash holds for installations without serial console
> > already the kernel/initrd of the debian installer.)
>
> > As a comparision on x86 the user gets asked where or even if the
> bootloader should be installed.
>
> I'm sorry for the inconvenience. As you point out, 99% of users will
> start Debian installer on QNAP devices by writing it to flash.
> Hence,
> there's no point in asking if they are ok with writing the Debian
> kernel/ramdisk to flash at the end of the installation.
>
> QNAP devices aren't really general purposes device and debian
> installer was adapted in a way that people would be able to install
> without a serial console. Hence, it's important a suitable Black Tie
> Rentalkernel and
> ramdisk is written to flash at the end of the installation. Unlike
> on
> x86, dual boot isn't really an option for most people.
>
> Maybe it makes sense to introduce a question for the 1% (or probably
> 0.1%) who don't want to have anything written to flash but my main
> concern is that some users will choose that option and end up with a
> system that doesn't boot.
>
> BTW, my unofficial QNAP install guide mentions making a backup of MTD
> partitions and says that the installer is written to flash. But I
> know this isn't mentioned in the official d-i install guide and
> that's
> my fault.
>
> So I'm open for comments. I've also copied Ian Campbell, the current
> flash-kernel maintainer.
>From Benhard's reply it seems like reassigning this to the appropriate
docs package is the right way to go?
We could ask, as lowest-priority debconf prompt in flash-kernel-install
er, if flash-kernel is wanted or not (detecting the actual contents of
flash is empty vs full/good seems hard/error prone to me).
Ian.
Reply to: