March 2005 - Anaconda-devel - Fedora mailing-lists

PATCH - Optimization to isomd5sum media check
by Dustin kirkland 26 Apr '05

26 Apr '05

Hi- I'm submitting a patch that provides an optimization to the isomd5sum media check functionality. Presently, the algorithm calculates a single md5sum after reading the entire disc, and reports a "PASS" or "FAIL" at that time. When there is corrupted data on the disc, a user will not know until the entire disc has been read--which can take quite a while. It would be optimal for the algorithm to exit the checking and report a failure as soon after reading the corrupted section as possible. I've accomplished this by adding two extra fields to the application data header of the ISO image. The ISO 9660 spec allows for a ~512-character string, of which were 224 characters are used. I've added two fields, "FRAGMENT SUMS" and "FRAGMENT COUNT", after which 326 characters are used. The "FRAGMENT COUNT" is an integer representing the number of fragments in which to break the entire ISO. The "FRAGMENT SUMS" is an ascii string of concatenated leading characters of md5sums of subsequent fragments. I've #define'd the length of FRAGMENT SUMS to 60 characters, as this number is evenly divisible by (2, 3, 4, 5, 6, 10, 12, 15, 20, 30). I've also #define'd FRAGMENT COUNT to be 20, which means that ISO will be broken into 20 fragments, each fragment represented by a 3-character string (the first 3 characters of the fragment's md5sum) in FRAGMENT SUMS. Looking at an example, the application data header used to look like: ISO MD5SUM = 1fca2222a947399d5b4a4226629be254;SKIPSECTORS = 15;RHLISOSTATUS=1;THIS IS NOT THE SAME AS RUNNING MD5SUM ON THIS ISO!! It now looks like: ISO MD5SUM = 1fca2222a947399d5b4a4226629be254;SKIPSECTORS = 15;RHLISOSTATUS=0;FRAGMENT SUMS = 67bd3263cccfda2d851a493b9a238e8877c45bca78e276e94437dd9f26e4;FRAGMENT COUNT= 20;THIS IS NOT THE SAME AS RUNNING MD5SUM ON THIS ISO!! The string [67bd3263cccfda2d851a493b9a238e8877c45bca78e276e94437dd9f26e4] is actually a concatenation of: md5(frag(0))[0-2] = 67b md5(frag(1))[0-2] = d32 md5(frag(2))[0-2] = 63c md5(frag(3))[0-2] = ccf md5(frag(4))[0-2] = da2 md5(frag(5))[0-2] = d85 md5(frag(6))[0-2] = 1a4 md5(frag(7))[0-2] = 93b md5(frag(8))[0-2] = 9a2 md5(frag(9))[0-2] = 38e md5(frag(10))[0-2] = 887 md5(frag(11))[0-2] = 7c4 md5(frag(12))[0-2] = 5bc md5(frag(13))[0-2] = a78 md5(frag(14))[0-2] = e27 md5(frag(15))[0-2] = 6e9 md5(frag(16))[0-2] = 443 md5(frag(17))[0-2] = 7dd md5(frag(18))[0-2] = 9f2 md5(frag(19))[0-2] = 6e4 Thus, if any one of these prescribed strings does not match what is calculated, the checking exits immediately, as a corruption has been detected. In the cases where the error is near the beginning of the disc, the time to identify the problem is greatly reduced. Note that there is a chance of collision. When the disc is broken into 20 fragments, each fragment is represented by a 3-character string, yielding a collision 1 in 16^3=4096 cases. Such a collision only means that the user will not benefit from the optimization, as the entire disc will be checked by the last operation just like it has always been done and the error will be detected at that time. One can reduce the chances of collision by decreasing the number of fragments, or expanding the length of the FRAGMENT SUMS string. This optimization occurs at virtually no performance cost. Media checking is overwhelmingly I/O bound, reading from disc. The additional calculations of smaller sums adds little additional processing time, as most of that time was being spent waiting on new data from disc. Please consider for inclusion in Anaconda, and I'll be glad to answer any questions or address concerns. I think the patch is pretty simple, and isolated to two files. Thanks! Dustin Kirkland

5 11

Updating HOWTO
by Gene Czarcinski 18 Apr '05

18 Apr '05

I have created a simple HOWTO for updating and recreating the Fedora Core 3 distribution CDs. It also includes information on creating DVDs instead of CDs as well as creating (and optionally expanding) the RescueCD. This document can be seen at http://czarc.net/HOWTO-Updating-fc3.html Before generally publishing this document, I thought some of the folks on this mailing list could comment and provide feedback to any errors I have made. I have not tried this with the FC-development (FC4) and would be interested in any differences which I could incorporate into the document. Gene

6 11

USB rescue image?
by Chris Adams 24 Mar '05

24 Mar '05

I've got diskboot.img on a USB stick and it is useful. It would be more useful to have a rescue image on there though; is it possible? It looks like anaconda (in FC3 anyway) doesn't know how to look at an unpartitioned USB device for stage2.img. Is there a trick to it? If I partition the USB flash, is there a way that I can make it bootable with syslinux? Would anaconda then recognize it as an install source (at least to get stage2.img for rescue mode)? -- Chris Adams <cmadams(a)hiwaay.net> Systems and Network Administrator - HiWAAY Internet Services I don't speak for anybody but myself - that's enough trouble.

1 0

Anaconda (10.x.x.x) Structure/Internals
by Shawn Starr 23 Mar '05

23 Mar '05

Hey, I'd like to know if there's any good WiKi's or HOWTOs on the Anaconda internals. I want to help modularize Anaconda by allowing users to create screens. One method I'm thinking about is the use of python plugins (in source form not DSO). Is anyone else working on a project? Or have any pointers? Thanks, Shawn.

1 0

Re: Re: CD Not Found
by D.Asensio 22 Mar '05

22 Mar '05

Thanks Gene. I've done what you told me. I've followed http://fedoranews.org/contributors/gene_czarcinski/update_distro/ to generate my own cd's but again the same error: 'CD Not Found' after selecting the language, keyboard and media. Does anyone has this error? or does anyone know how to solve it? Thanks again.

1 0

buildinstall error
by D.Asensio 21 Mar '05

21 Mar '05

I'm trying to use anaconda in my distribution and when I try to run the buildinstall script I get the following output: """ Running buildinstall... /anaconda/base/fc3/i386/buildinstall.tree.5041 /anaconda/base /anaconda/base Going to run buildinstall again Building images... Assembling package list... Expanding text packages... Expanding graphical packages... retrieving timezones argument is not an RPM package cpio: premature end of archive ERROR: Unable to expand timezone data () Aborting instroot creation... Getting package order... Traceback (most recent call last): File "/tmp/treedir.5054/instimage/usr/lib/anaconda-runtime/pkgorder", line 12, in ? from anaconda_log import anaconda_log ImportError: No module named anaconda_log Making images... /tmp/treedir.5054/instimage/usr/share/hwdata/pcitable doesn't exist /tmp/treedir.5054/instimage/usr/share/hwdata/pci.ids doesn't exist /tmp/treedir.5054/instimage/usr/lib/anaconda/lang-table doesn't exist Writing .discinfo file timestamp not specified; using the current time """ I'm checking the code of the script to try to find the what's wrong, but still no result. The error is in the execution of upd-instroot which is q really big script. I thing the first error is produced because I don't install some package which is neccesary ("argument is not an RPM package"). So, I would like to know what packages I should install always I want to use anaconda. Nowadays I have anaconda-runtime, is anaconda package also neccesary? Thanks.

1 0

CD Not Found
by D.Asensio 21 Mar '05

21 Mar '05

I have followed the doc written by Gene Czarcinski at http://czarc.net/HOWTO-Updating-fc3.html and when I try to install my distribution I get a 'CD Not Found' screen after selecting my language and keyboard... I have tried to insert all the CD's generated but I always get that error. What could I have done wrong? How can I solve it?

2 1

rpmbuild make error "smp.c attribute ignored"
by abayomi onibudo 17 Mar '05

17 Mar '05

Hi, i have just mdofied a custom anaconda src such that it only allows installation for desktop-install i.e. server, custom and workstation sources taken out of BUILD/anaconda-10.0/installclassses/ I have also created the patch using diff -ruN anaconda-9.2.orig anaconda-9.2 > /usr/src/redhat/SOURCES/my-changes.patch and updated the anaconda.spec with custom patch can anyone tell me why i am getting the follow rpmbuild error error smp.c:228: warning: `__packed__' attribute ignored smp.c:245: warning: `__packed__' attribute ignored smp.c:256: warning: `__packed__' attribute ignored make[1]: *** [smp.lo] Error 1 make[1]: Leaving directory `/usr/src/redhat/BUILD/anaconda-10.0/isys' make: *** [subdirs] Error 1 error: Bad exit status from /var/tmp/rpm-tmp.91765 (%build) RPM build errors: Bad exit status from /var/tmp/rpm-tmp.91765 (%build) ---- thanks Bayo

3 3

How to change the text Messages and GUI screens while Installation
by johnpaul 17 Mar '05

17 Mar '05

Hi I am creating a customized CD for Installing RedHat 9. While installing in the text mode after running the loader during package installation I want to change the copyright message. In the anaconda source.I have even found the file ( text.py)(Hope I am right). I have created the RPM again by changing that file. I a have replaced the loader,init of the initrd.img with the new one from the RPM created. But again while installing I am getting the same old copy right message.Is there any steps to change the initrd.img after changing the python scripts in the Anaconda. Also give me some inputs to rebuild the GUI images . please help me Thanks John -- This message was scanned for spam and viruses by BitDefender. For more information please visit http://linux.bitdefender.com/

2 1

Pre-release / Beta
by Jaswinder Singh Kohli 16 Mar '05

16 Mar '05

Hi, I am trying to build a custom installation of Fedora Core 3. All is going well until i was using the kickstart method. But recently i tried to install it in interactive mode and got two strange messages. Thus a unsuccessfull install. The Messages are... 1. After the first boot into graphical interface and clicking next ... Something like " Thankyou for downloading this pre-release version of Fedora Custom Core 3,. This is not final version......" etc... and 2. After all the phases like, parition, boot loader, time zone selection when the message about scanning packages comes. the system says .. " You are trying to install on a machine which isn't supported by this release of Fedora Custom Core 3" Any details, about my what i am doing wrong... Thanks a lot in advance... -- Regards JSK jskohli (AT) gmail (DOT) com

1 0

Ejecting boot.iso (again)
by Paul Nasrat 16 Mar '05

16 Mar '05

It seems as if we unlink /tmp/cdrom so eject fails :( Looking at devMakeInode we already unlink the path so the following should mean that we can eject in the error conditions reached with boot.iso whilst preserving current behaviour when looping through the cdroms. Views? Paul

1 0

RE: 2.1AS kickstart - manual network config CD/DVD install
by Andrew Rechenberg Lists 14 Mar '05

14 Mar '05

Patrick, Thanks for the speedy reply. > > Take a look at setInstallData() in kickstart.py. > self.readKickstart() gets called twice, once before %pre and > once after (or at least it does so in RHEL 3). > The kickstart.py in RedHat/instimg/usr/lib/anaconda on a RHEL2.1ASU5 disc only calls self.readKickstart() once :\ I guess I could patch my kickstart.py to read it twice. I'll check that file on 3 or later and modify mine accordingly to see if that works :) > > > In my ks.cfg I have "network --bootproto dhcp --device > eth0" and then > > in the %pre section I prompt for network configuration > information and > > then use sed to change the "network" line, redirect the output to a > > temporary file and then copy this temp file back to /tmp/ks.cfg. > > Why not take the network setting out of your kickstart file > and have anaconda prompt for it with it's own network screen? > >From what I've read in other threads, when doing a CDROM-based kickstart on distros before RHEL3, anaconda doesn't care if "network" is missing and uses DHCP automatically and doesn't prompt. I can certainly try it though :) I'll try that first since I don't have to mess with modifying anaconda. Thanks for your help, Andy. Confidentiality Notice: This e-mail message including attachments, if any, is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message. If you are the intended recipient, but do not wish to receive communications through this medium, please so advise the sender immediately.

2 1

2.1AS kickstart - manual network config CD/DVD install
by Andrew Rechenberg Lists 14 Mar '05

14 Mar '05

Good day, I've been doing some research on prompting for manual network settings in the %pre section of a kickstart file and it appears that anaconda isn't behaving as some people think that it should. According to a couple of threads (I believe on) the kickstart-list and/or the anaconda-devel-list, the ks.cfg file is supposed to be re-read after the %pre section script is executed. On a RHEL2.1ASU5 install, this re-read doesn't seem to be happening. In my ks.cfg I have "network --bootproto dhcp --device eth0" and then in the %pre section I prompt for network configuration information and then use sed to change the "network" line, redirect the output to a temporary file and then copy this temp file back to /tmp/ks.cfg. Here are some of the threads I've checked out: https://www.redhat.com/archives/kickstart-list/2004-March/msg00075.html http://www.redhat.com/archives/anaconda-devel-list/2003-June/msg00149.ht ml (this thread indicates ks.cfg is re-read) So, my question is: Is ks.cfg re-read after the %pre section? What could be my problem if it is? I would like to get the network configuration in the %pre section so the installer can answer the questions and then move on to the next install. Thanks for your help, Andy. Regards, Andrew Rechenberg Infrastructure Team, Sherman Financial Group Confidentiality Notice: This e-mail message including attachments, if any, is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message. If you are the intended recipient, but do not wish to receive communications through this medium, please so advise the sender immediately.

2 1

Re: anaconda with lvm default
by Michael Honeyfield 12 Mar '05

12 Mar '05

> On Sat, Mar 12, 2005 at 02:49:45AM +1300, Michael Honeyfield wrote: > > I am just curious the reasoning behind using lvm as the default > > partitioning method in the anaconda version shipped with FC3. > > LVM is good. On a desktop PC with a single disk? Makes little sense. Not disputing the fact that lvm is good or bad, asking for the reasoning to making it the default partitioning method. Cheers Mike

4 4

anaconda with lvm default
by Michael Honeyfield 11 Mar '05

11 Mar '05

I am just curious the reasoning behind using lvm as the default partitioning method in the anaconda version shipped with FC3. I have been looking through the change logs and past mailing lists, but I have not seen any reasons or comments indicating the whys... Any comments would be great Thanks Mike

4 3

Re: anaconda with lvm default
by Michael Honeyfield 11 Mar '05

11 Mar '05

> > I am just curious the reasoning behind using lvm as the default partitioning method in the anaconda version shipped with FC3. > > > > I have been looking through the change logs and past mailing lists, but I have not seen any reasons or comments indicating the whys... > > Well I got to this party after the decision to use LVM for everything > had been made, but I can share my reasons for why I think it's a good > idea. > > First, it frees us from a lot of the disklabel constraints. We don't > really have to worry about four partition limits and extended partitions > and all that junk if we can just make one big LVM blob and cram a bunch > of partitions in there. So that makes it more flexible for people who > want the really crazy partitioning schemes. > > Second, it's nice to have fewer cases to deal with in the partitioning > code. If we know everyone's got LVM, it means there's less we have to > take into account. Of course right now, everyone doesn't have LVM. But > hopefully years down the road, everyone will. :) > > There are probably more but these are the two that immediately popped > into my head. Personally, I find LVM to be more flexible and getting > used to it now on my own machines means I'll know how to use it when I > finally get that giant disk array set up at home. Thanks for that Chris... I have been reading a bit more about it (my previous usage of lvm has be nil) and can see its strong points. I have more questions about lvm, but I think they are OT on this list. Thanks again Mike

1 0

Anaconda Bug?
by Ronald Reed 11 Mar '05

11 Mar '05

Anaconda does not appear to follow the DEVICE option in the ks.cfg file. I have the folling lines in my syslinux.cfg: label nfs-dell kernel vmlinuz append expert noprobe ks=cdrom:/nfs-dell-ks.cfg initrd=initrd.img In the nfs-dell-ks.cfg file I have the following: lang en_US.UTF-8 langsupport --default en_US.UTF-8 en_US.UTF-8 network --device eth0 --bootproto dhcp --hostname nfs-basic firewall --disabled cdrom device scsi aacraid device scsi megaraid device scsi aic7xxx device eth eepro100 keyboard us bootloader --location=mbr clearpart --linux --drives sda part /boot --fstype "ext3" --size=128 --ondisk=sda part swap --size=512 --ondisk=sda part / --fstype "ext3" --size=100 --grow --ondisk=sda install mouse genericwheelps/2 --device psaux timezone --utc Etc/GMT rootpw --iscrypted ### authconfig --enableshadow --enablemd5 skipx reboot >From what I understand, the "expert noprobe" that is appended to the boot prompt makes anaconda skip probeing for devices. The device options in the Kickstart file are supposed to define the devices. This doesn't work. Am I wrong in how this is supposed to work? -- =========================== Ron Reed RedHat Certified Engineer SGP Computer Department Manager Unix Systems Administrator ARM SGP CART Site (580)388-4053 ron.reed(a)arm.gov

2 1

Kickstart in RHEL 3 goes interactive
by Brian Long 09 Mar '05

09 Mar '05

Hello, I would like to know if Jeremy or anyone else has considered an alternate failure scenario when a kickstart is supposed to be completely scripted and it goes interactive because the NFS server is not found. Right now, if someone boots and they're unable to reach the NFS server (either because IP is misconfigured, or the host has dual NICs, or whatever), the installer goes into interactive mode and asks the Language, Keyboard and the NFS server location. I'd like to define an alternate failure scenario that just displays a message: "Your installation is unable to complete properly. This is most likely due to a network issue. Please contact your support organization for further assistance." When I have hundreds of folks installing linux via kickstart and the first thing they see asks them for the language, they assume they're supposed to answer it. I'd like to make sure they never get the interactive questions; instead it should just fail with a message I define. Any ideas on how to accomplish this and if it makes sense to define a new ks.cfg parameter that enables this new behavior? /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s

1 0

RE: kernel
by Oden, James 08 Mar '05

08 Mar '05

> Hello, > > Is it possible to replace the boot kernel on the CD with one that is > compiled with modifications, may be with the same kernel version number, and then > copied to the cd. Has anyone tried this ? Several have. There is a script out there to help with this. If you can search the list you should be able to find it. Basically, there are two things you have to update: - the initrd.img that contains the kernel itself. - the various module balls that contain the modules. There are also two strategies: - rebuild the entire iso using anacondas normal build tools. - extract anaconda images, place new kernel and modules in place and update dep files. Cheers...james

1 0

kernel
by Sameer Kamat 08 Mar '05

08 Mar '05

Hello, Is it possible to replace the boot kernel on the CD with one that is compiled with modifications, may be with the same kernel version number, and then copied to the cd. Has anyone tried this ? Thanks, Sameer. Internal Virus Database is out-of-date. Checked by AVG Anti-Virus. Version: 7.0.300 / Virus Database: 266.5.7 - Release Date: 3/1/2005

1 0

Where is "Red Hat Linux" image?
by bonzaipanthera 07 Mar '05

07 Mar '05

2 2

custom rh7.3 GUI startup failed.. Pls help...
by bonzaipanthera 07 Mar '05

07 Mar '05

2 4

anaconda doesn't install kernel-smp
by Daniel Spratlen 03 Mar '05

03 Mar '05

I'm working on a customized install disk for our x86-64 systems using RHEL ES 3.0 update 4. My problem is that even though the smp kernel is on the CD, and even though the systems are dual cpu systems, anaconda won't install the smp kernel. I can manually install the smp kernel after I've booted the freshly installed system and everything works as intended. What could I be missing that's causing anaconda to skip installing the smp kernel? thanks, Daniel

2 2

--rootpath questions/barfs
by Armijn Hemel 02 Mar '05

02 Mar '05

hi, I'm playing around a bit with Anaconda. I want to use it as an installer for custom UML images and the likes. Anaconda barfed on me and I'm not really sure what to do to fix it. What I did was the following: - start Xnest on localhost:2 (without access control) - export DISPLAY=localhost:2 - # anaconda --graphical --rootpath=/tmp/anaconda --method=cdrom:///dev/hdb:/tmp/source where /dev/hdb has my Fedora Core 3 DVD, and which is mounted on /tmp/source Everything works fine, miniwm and everything starts, I can configure the firewall, timezone, network interfaces, packages, etc. When I start installing Anaconda barfs with: IndexError: list index out of range The complete traceback: Traceback (most recent call last): File "/usr/lib/anaconda/gui.py", line 1128, in handleRenderCallback self.currentWindow.renderCallback() File "/usr/lib/anaconda/iw/progress_gui.py", line 242, in renderCallback self.intf.icw.nextClicked() File "/usr/lib/anaconda/gui.py", line 872, in nextClicked self.dispatch.gotoNext() File "/usr/lib/anaconda/dispatch.py", line 173, in gotoNext self.moveStep() File "/usr/lib/anaconda/dispatch.py", line 241, in moveStep rc = apply(func, self.bindArgs(args)) File "/usr/lib/anaconda/packages.py", line 759, in doPreInstall if method.systemMounted (id.fsset, instPath): File "/usr/lib/anaconda/image.py", line 183, in systemMounted self.loopbackFile = "%s%s%s" % (chroot, IndexError: list index out of range The method where it barfs: def systemMounted(self, fsset, chroot): print "chroot: ", chroot print fsset.filesystemSpace(chroot) self.loopbackFile = "%s%s%s" % (chroot, fsset.filesystemSpace(chroot)[0][0], "/rhinstall-stage2.img") With a few debug statements I've verified that fsset.filesystemSpace(chroot) is [], which of course causes the IndexError. My questions: what does this filesystemSpace actually do and how can I fix this particular problem, I know it's a bit of an exotic use of Anaconda, but it would be neat to be able to use it this way :) Oh, I'm using anaconda 10.2.0.24 on an up to date Fedora Core 3 system. thanks in advance, armijn -- --------------------------------------------------------------------------- armijn(a)uulug.nl | http://www.uulug.nl/ | UULug: Utrecht Linux Users Group ---------------------------------------------------------------------------

1 0

[PATCH] empty the blacklist
by Bill Nottingham 02 Mar '05

02 Mar '05

Anyone uprading from a release with the old broken linuxconf packages (i.e., 7.2 or earlier) is almost certainly going to enough other problems. Fixes 142893, incidentally. Bill

2 1

remove minimum and full install
by Vincent FONTENEAU 01 Mar '05

01 Mar '05

Hi, I'm using FC3 and i've completely change comps.xml file. I've got my own installation for the distribution. I'm booting with the new installation cdrom and then I could choose all options i want. At the end i'm going with packages selection and i always got the options to install the minimum packages or perform full installation in "diverts" options. is it possible to remove those two options or bye the way to remove the menu "diverts" from the package selection Thank Vincent

2 2

Problems with Compat Arch Support group
by Brian Long 01 Mar '05

01 Mar '05

I've having problems with the "Compatibility Arch Support" group on x86_64 kickstarts. We generate comps.xml from scratch and our Compat Arch Support group contains all the biarch RPMs and the "biarchonly" flag is set to true (the group looks the same as the stock comps.xml). The problem is the biarch RPMs are not getting installed in the proper order. Sometimes the i386 RPMs get installed after the x86_64 RPMs, so they take precedence (and break the system). How can I debug this further? Is there somewhere I can instrument Anaconda's code to figure out the problem? This happens with RHEL 3 Update 2, 3 and 4. Thanks. /Brian/ -- Brian Long | | | IT Data Center Systems | .|||. .|||. Cisco Linux Developer | ..:|||||||:...:|||||||:.. Phone: (919) 392-7363 | C i s c o S y s t e m s

4 5

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Anaconda-devel March 2005