Christoph Wickert wrote:
Am Dienstag, den 14.07.2009, 09:27 -0600 schrieb Douglas McClendon:
Douglas McClendon wrote:
Christoph Wickert wrote:
Am Dienstag, den 14.07.2009, 08:09 -0600 schrieb Douglas McClendon:
2. Imagine after the installation you switch rebootless to the new
system and install a kmod. But you are still running the kernel
from the installation medium and kmods get installed for the
running kernel, which not necessarily needs to be the one that
was installed.
As with a current LiveOS installation, the installation media kernel is
the running kernel. (unless the f11 installer already allows you to
trigger a chrooted yum update as part of install).
Ok, I'll show my good fedora developer maturity and call it a 'night'
now that I'm starting to get sloppy. That should have been worded-
As with a current LiveOS installation, the installation media kernel is
the running kernel. Even if the f11 installer already allows you to
trigger a chrooted yum update as part of the install, you won't be
running the updated kernel until after a reboot.
There is no chrooted yum update but the updated packages get installed
*instead* of the old ones from the installation media.
I believe you are speaking of the most traditional (DVD) installer, and
not the LiveOS/CD/USB installer. I just added the following to the
feature wiki-
clarification note: This is only an alternate method to the current
LiveOS installer. This method does not provide any interesting
alternative to the old school non-live DVD installer. Though it does
provide an alternative for DVD-sized LiveOS spin's installers. This
architecture could provide a rebootless equivalent to what it sounds
like opensuse achieves with kexec, but that is just an idea with no
proof of concept yet, unrelated to this feature. Nor does this feature
pertain in any way to upgrade scenarios, just as the current Fedora
LiveOS installer does not (I believe) support any kind of upgrade. (I
believe vanilla anaconda may be usable to perform upgrades from the
current LiveCD, but that is not a use-case that is well
documented/advertised. Wiki editors, please confirm/deny)
Thanks everyone for the vetting so far.
Thanks a lot for the proposal and the work you've put into it. I'm sure
we all are interested in it, but many of use are just a little skeptical
if it really works out. Please don't let this skepticism scare you. :)
No need to worry, I've been around the fedora-devel block once or twice
before. The technique/technology here I first published as a bash
script on this list a couple years ago. I think what is sinking in is
that people need GUIs and lots of clear documentation in order to
mitigate their skepticism. It's getting there...
Thanks,
-dmc
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list