archive-installer backend

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Yesterday, I dove into anaconda for the first time ever (somehow I
skipped touching anaconda in all my time at Red Hat) and implemented
a new backend that installs system images from tarballs (later,
potentially also cpio archives).  My intent is that the tarballs
can be on the media or fetched over the network (NFS, http, ftp),
but for now I've only tested installing from a single archive file
on the media.  This seems interesting to me for replicating systems,
and for pre-staged installs of a single image.

I made a tarball of a minimal (text) F15 installation, dropped it
onto F15 DVD media with only my backend-related changes applied to
the second stage tree, and then installed it.  The installed
system works fine (as long as I remember to use --selinux when
creating the tarball).

To try to compare apples to apples, I did two installs in text
mode using identical media on the same exact hardware (ThinkPad
T60, 320GB 7200 RPM drive, 3GB RAM); one with the F15 DVD x86_64
media, the other using the same iso image modified only by adding
my backend, removing the packages, and including the tar archive
instead.

I timed (by stopwatch) from accepting writing to disk through (a)
starting package installation proper (not the previous prep step
that just doesn't exist for the archive case), (b) 100% of packages
install step, and (c) reboot prompt.  (This meant that my timings
included the process of creating LVs and filesystems; I didn't
think I could time as accurately otherwise.

            (a) start inst  (b) 100%        (c) reboot prompt
packages:   1m00s           3m15s           3m53s
archive:    0m26s           1m00s           1m40s

I expect that when I move on to testing NFS I'll see a more
substantial difference because right now it appears that archive
install is speed-limited by the DVD drive on the test machine.

Now that I have a working prototype, I thought I'd ask whether this
is something that would be considered for upstream official anaconda.

_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list


[Index of Archives]     [Kickstart]     [Fedora Users]     [Fedora Legacy List]     [Fedora Maintainers]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [Yosemite Photos]     [KDE Users]     [Fedora Tools]
  Powered by Linux