Anaconda-devel August 2012

anaconda-devel@lists.stg.fedoraproject.org

29 participants
32 discussions

Ability to install to loopmounted filesystem - mentor wanted
by Tirtha Chatterjee 02 Dec '12

02 Dec '12

Hello. I am trying to implement something like Wubi on Fedora. For this, I first need to make it possible to install to loopmounted filesystems through anaconda. I don't think a lot of modification to the existing source code is required. I tried going through the source code, but it seemed quite big to manage without some guidance. I would be extremely happy if someone guided me on this. Hoping for an affirmative reply :-) -- Regards Tirtha Chatterjee KDE developer http://wyuka.co.cc/

2 3

console handling with tmux
by Will Woods 06 Sep '12

06 Sep '12

So now that we have tmux handling consoles for us, the rules for console stuff have changed a little. Basically, anything in anaconda that deals directly with VTs or /dev/tty* should be considered deprecated. tmux has a fairly robust command set that will let you manipulate its state (and thus what's shown on the user's console). For example, here are some common tasks and possible replacements: Old and busted: Sending log output to tty{3,4,5...} New hotness: Send log output to log files; tail log in tmux window Old: Run debugger in a new VT (e.g. tty6) New: Run debugger in a new window: `tmux new-window -n pdb "[debugger-command]"` Old: Activate (switch to) debugger on tty6 New: Not needed (new-window automatically switches) Old: Switch back to tty1 using isys.vtActivate New: Switch back to first window: `tmux select-window -t anaconda:1` (Note that when you're specifying a window by hand, you can probably get away with just `tmux selectw -t 1`, but in scripts inside anaconda we probably want to specify the full [session-name]:[index]) For more details, consult the man page. Here's a link to it on the web: http://www.openbsd.org/cgi-bin/man.cgi?query=tmux&sektion=1 -w

5 6

Upating variables from a %pre script
by Matt Bryant 30 Aug '12

30 Aug '12

All, I was wondring if someone could point me in the right direction Is it possible to update a vaiable in kickstart file give user input in a %pre script. Basically i have a varibale buildType and whilst I can override the default value of the profile using ksmeta input when adding a system to cobbler would like to also be able to override given user input from %pre script. Have tried just about everything i can think of including creating a file ks.settings which held the variations #set global $buildType = "value" #set global buildType ="value" #set buildType = "value" #set $buildType = "value" then doing a #include but that doesnt seem to work ... Is there in fact a way this can be done ?? Any help appreciated. rgds Matt B.

2 3

Handling updates and product images
by Jesse Keating 24 Aug '12

24 Aug '12

With F18 we have a wrinkle with handling of updates and product images. Previously in dracut we would unpack them and then overlay them on top of the newroot (/sysroot/). To make this work, our updates and product images would have a path within them, /tmp/updates/ or /tmp/product/. This worked fine, until /tmp became a tmpfs mount created by systemd during system bring up. When this mount happens it will cover up anything we've put in /tmp/ during dracut and thus we'll lose our updates. The quick and easy fix we've gone with for F18 Alpha is to mask the tmpfs service so that it won't run. This brings things inline with how it worked in F17. Will has expressed concern that we shouldn't deviate from the system like this and instead work around it, so I've explored some options. There is no order to these options. Option A) Make use of an init service that copies everything from /run/initramfs/state/ onto the / filesystem during bootup. To accomplish this we would have our updates image use /run/initramfs/state/tmp/updates (and /run/initramfs/state/product) as the location of the files. Provided the fedora-import-state service runs after the tmpfs mount, we will have our content in /tmp/updates,product by the time anaconda runs. Pros: small amount of code change, no change in anaconda code for dealing with update content. Cons: possible ram overusage. Copying files from one ramfs to a tmpfs may double the amount of ram the files are using. scripts/makeupdates | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) Option B) Create a symlink from /run/initramfs/inst.{updates,product} to /tmp/{updates,product} when anaconda launches. This option still makes use of the /run/initramfs/ pass through, but instead of relying on a service to copy files into / we'll just make a symlink early in the anaconda process (right at the top of when we'll deal with the python updates). Pros: still a small amount of code change, and potentially reduces memory footprint vs option A. Cons: It feels like a hack on top of a hack. anaconda | 6 ++++++ scripts/makeupdates | 7 ++++--- 2 files changed, 10 insertions(+), 3 deletions(-) Option C) Just directly use update content from /run/initramfs/inst.{updates,product}. Don't rely on a service to copy, nor make a crutch symlink, just directly use the content from those directories. This requires the most code change, and even breaks running tests as non-root (as there is a test that creates a bogus product .buildstamp and that fails if you're creating it in /run/ as non-root). Pros: It's not a crutch or a paperover, it's a direct use. Doesn't rely upon a service out of our control or timing issues. Cons: Lots of code churn, breaks tests as nonroot. anaconda | 16 ++++++++-------- data/liveinst/liveinst | 12 ++++++------ pyanaconda/constants.py | 2 +- pyanaconda/installclass.py | 4 ++-- pyanaconda/localeinfo.py | 4 ++-- pyanaconda/packaging/yumpayload.py | 8 ++++---- pyanaconda/product.py | 2 +- pyanaconda/ui/gui/__init__.py | 2 +- pyanaconda/ui/gui/hubs/progress.py | 4 ++-- pyanaconda/yuminstall.py | 2 +- scripts/makeupdates | 9 +++++---- tests/pyanaconda_test/product_test.py | 4 +++- 12 files changed, 36 insertions(+), 33 deletions(-) Option D) Keep masking the tmpfs mount. Pros: very simple change in lorax, no other code changes. No file duplication in ram. Cons: Potential to overflow the overlay, cannot be swapped to disk, diverges from Standard System. My personal preference would be option D, because it involves the least amount of code change and is the most simple solution. Secondary to that I'm leaning toward option D, again little code change. I'm posting here for more opinions, perhaps options I haven't thought of. -- Jesse Keating Fedora -- Freedom² is a feature!

2 1

running livemedia-creator with 1GB
by David A. Marlin 23 Aug '12

23 Aug '12

I am running on F17 and using: anaconda-17.29-1.fc17 lorax-17.26-1.fc17 I was creating a test image using livemedia-creator on a resource constrained system (only 1GB memory) and encountered an error. "You have not specified a swap partition. 917 MB of memory is required to continue installation without a swap partition, but you only have 898 MB." But I have a large swap partition on the build system, and it is active. I have successfully created images before, but this time I used a new kickstart that did not include a 'swap' partition definition. clearpart --all part /boot --size 255 --fstype vfat --asprimary --label=boot part / --size 2000 --fstype ext4 --label=rootfs If I add a small swap partition in the kickstart, it works. Is this the intended behavior? Is there a way to create a target image that does not include a swap partition? Thank you, d.marlin

2 4

0-day exploit in anaconda
by John Reiser 21 Aug '12

21 Aug '12

Last night and this morning, I ran Fedora 18-Alpha-TC3 netinst.iso (x86_64) burned to DVD, and met the bug https://bugzilla.redhat.com/show_bug.cgi?id=849211 . About 40 minutes ago, I re-ran the DVD but the bug had been fixed. I never explicitly approved any Fedora 18 package signing key. I believe that none (zero) of the current "Fedora 18" [rawhide] packages have been signed. The download files from directory link http://dl.fedoraproject.org/pub/alt/stage/18-Alpha-TC3/Fedora/x86_64/iso/ are not accessed by the secure protocol https:// . There is a *-CHECKSUM given, but it is not signed, either. The fix for the bug 849211 was automatically downloaded and installed, insecurely. That's just a short step away from a 0-day exploit in the installer. --

3 4

Re: partitioning expectations for F18
by Chris Murphy 20 Aug '12

20 Aug '12

I asked this on the test list, but it's more appropriate here. What are the expected partitioning states for F18? Including Apple hardware, I'm imagining roughly seven possible states (MBR vs GPT, btrfs vs LVM), none of which are what Alpha TC3 produces. What I get is no LVM usage, an extended partition just for /home, and ext4. Is this transitional behavior? Chris Murphy

5 7

expectations for Fedora 18-Alpha-TC3
by John Reiser 17 Aug '12

17 Aug '12

How much of Fedora 18-Alpha-TC3 netinst.iso should I expect to work? It booted from DVD on bare metal and brought up graphics, but after that I encountered many problems. Media_check did not run, despite being included in the default selection on the boot screen ("verify source..."). <Tab> did not move the focus to the next graphical button on the Choose Language screen. There was no progress indicator for Installation Source, Software Selection (downloading repo info, checking dependencies), or Installation Destination (examining storage devices). The left box (Desktop) of Installation type (Base, GNOME, KDE, LXDE, Sugar, XFCE) was unclear. Is it a radio-button list (select only one) but without the buttons? The box abuts the left margin, and looks like something has been cut off on the left. The right box (Addons) was hard to track. There should be a separate list of current selections. There should be an option to hide/display the descriptions (with the names always displayed.) It should be possible to search by regex. Display of dependencies should be available. Installation Destination gave only the barest info: disk manufacturer and model, with gross capacity. There was no connection information (sda, sdb, ...; pci-...-scsi-...; usbN; etc.). Horizontal scrolling sucks; my box has 5 drives, and I want to see them all at once. Sometimes I have multiple drives of the same make and model, so I need to see the serial numbers, too. I want to see the volume label type (MSDOS, GPT, Apple, ...), plus the partitions, size, label, filesystem type, used space, unused space, last mount point: everything that gparted displays. There was no message "No LVM2 structures were found" which is required in order to increase my confidence and help detect errors. Then the installer crashed, and the bug report via network to bugzilla was incomplete because "you do not have permission to access the file". [Bug 848994] New: IndexError: list index out of range --

8 20

Re: data/Makefile.am data/linuxrc.s390
by Denise Dumas 17 Aug '12

17 Aug '12

Congratulations! Break out the champagne and party like its 1999 - uh wait ... Denise On 08/16/2012 05:29 PM, David Cantrell wrote: > data/Makefile.am | 7 > data/linuxrc.s390 | 3079 ------------------------------------------------------ > 2 files changed, 1 insertion(+), 3085 deletions(-) > > New commits: > commit 011ea0a1779459ed20990ddf52166aa75a9c1382 > Author: David Cantrell<dcantrell(a)redhat.com> > Date: Thu Aug 16 17:27:13 2012 -0400 > > Remove linuxrc.s390 > > The mainframe no longer needs an overly complicated shell script to > bring up the complicated installation environment. It boots the way > other architectures do, except on punch cards and lab coats. > >

2 1

Bill's patch for software selection + comps metadata
by Máirín Duffy 17 Aug '12

17 Aug '12

For those of us not on anaconda-patches list (like myself and the other designers I know are monitoring this list :) ): https://lists.fedorahosted.org/pipermail/anaconda-patches/2012-July/000165.… Here is the screencast Bill Nottingham made of the UI changes in the patch: http://notting.fedorapeople.org/installer.webm Some things Chris, Bill, and I discussed about the patch: - Overall a great idea. - 'Installation Type' is maybe an odd term to use. We talked about using the term 'Environment' instead - that term is inclusive of all the desktop options and (no-desktop) minimal install that are in that selection box. - We use the term 'install class' a lot, so it might be better to pick a different term. Bill has used install class here to mean a desktop + the comps groups that are compatible with it = a base package set. Some alternatives Bill proposed (with some thesaurus help): selection collection combination compilation assembly mix My suggestion is to use "environment collection" (even though it's long) because I think collection is the most appropriate term (a group of sets, think Flickr 'collections' vs 'sets') but it's also important to use the term 'environment' in it so that someone looking at the comps knows that it relates to the 'environment' widget in the UI. Thoughts? ~m

4 7

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Anaconda-devel August 2012