Anaconda-devel March 2005

anaconda-devel@lists.stg.fedoraproject.org

31 participants
27 discussions

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

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