Anaconda-devel September 2008

anaconda-devel@lists.fedoraproject.org

33 participants
41 discussions

Various iSCSI fixes [PATCH 2/4]: anaconda-driveIsIscsi.patch
by Hans de Goede 13 Sep '08

13 Sep '08

Hi all, On newer kernels the symlinks are arranged differently, causing driveIsIscsi to fail to detect iscsi drives in current rawhide. This patch fixes this, fixing rh 461839, rh 461841. Regards, Hans On newer kernels the symlinks are arranged differently, causing driveIsIscsi to fail to detect iscsi drives in current rawhide. This patch fixes this, fixing rh 461839, rh 461841. diff --git a/isys/isys.py b/isys/isys.py index 88a6987..8cec02a 100755 --- a/isys/isys.py +++ b/isys/isys.py @@ -893,7 +893,7 @@ def driveIsIscsi(device): # ewww. just ewww. if not os.path.islink("/sys/block/%s/device" %(device,)): return False - target = os.readlink("/sys/block/%s/device" %(device,)) + target = os.path.realpath("/sys/block/%s/device" %(device,)) if re.search("/platform/host[0-9]*/session[0-9]*/target[0-9]*:[0-9]*:[0-9]*/[0-9]*:[0-9]*:[0-9]*:[0-9]*", target) is not None: return True return False

1 0

Various iSCSI fixes [PATCH 1/4]: anaconda-rh454135.patch
by Hans de Goede 13 Sep '08

13 Sep '08

Hi All, The attached patch only allows selection of devices known by booty as bootloader target disk, fixing the crash reported in bug 454135 Regards, Hans The attached patch only allows selection of devices known by booty as bootloader target disk, fixing the crash reported in bug 454135 diff --git a/iw/autopart_type.py b/iw/autopart_type.py index c1366f7..714d04a 100644 --- a/iw/autopart_type.py +++ b/iw/autopart_type.py @@ -365,6 +365,8 @@ class PartitionTypeWindow(InstallWindow): else: defaultBoot = None for disk in self.diskset.disks.values(): + if not disk.dev.path[5:] in self.anaconda.id.bootloader.drivelist: + continue size = partedUtils.getDeviceSizeMB(disk.dev) dispstr = "%s %8.0f MB %s" %(disk.dev.path[5:], size, disk.dev.model) i = bootstore.append(None)

1 0

Various iSCSI fixes [PATCH 0/4]
by Hans de Goede 13 Sep '08

13 Sep '08

Hi All, Here is the first bunch of fixes from all the iSCSI work I've been doing I'm sending these to the list as I don't have git access yet (this is being worked on). I don't know if I'll get around to committing these to the master before I leave for plumbers, so I would appreciate it if someone could commit them to master for me to make sure they make the F-10 rc's See the messages with the actual patches for a longer description of each patch anaconda-rh454135.patch: -fix bug 454135 anaconda-driveIsIscsi.patch: -fix anaconda not seeing iscsi as iscsi in rawhide anaconda-iscsi-rhel5-to-master1.patch: -port all rhel5 iscsi changes done before Fudcon to master anaconda-iscsi-rhel5-to-master2.patch -port all rhel5 iscsi changes done after Fudcon to master Thanks & Regards, Hans

1 0

[PATCH] Fixup DiskSet._askForLabelPermission()
by Mark McLoughlin 12 Sep '08

12 Sep '08

This function seems to have gotten horribly screwed up lately: - We never actually query the user about initialising the partition table; so if you've got an unitialized partition table, there's no way to install - We've create two versions of the message string and use neither - We test for "ks and (drive in clearDevs) and initAll" twice; the second test can never be true, obviously - We test for "not intf" twice, again the second can't be true Patch is untested, but it's clearly an improvement :-) Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- partedUtils.py | 36 +++++++++--------------------------- 1 files changed, 9 insertions(+), 27 deletions(-) diff --git a/partedUtils.py b/partedUtils.py index 725ccc5..17a1208 100644 --- a/partedUtils.py +++ b/partedUtils.py @@ -1099,41 +1099,23 @@ class DiskSet: return False rc = 0 - if ks and (drive in clearDevs) and initAll: + if (ks and (drive in clearDevs) and initAll) or \ + self.isDisciplineFBA(drive): rc = 1 - else: - if not intf: - self._removeDisk(drive) - return False - msg = _("The partition table on device %s was unreadable. " + elif intf: + deviceFile = "/dev/" + drive + dev = parted.PedDevice.get(deviceFile) + + msg = _("The partition table on device %s (%s %-0.f MB) was unreadable.\n" "To create new partitions it must be initialized, " "causing the loss of ALL DATA on this drive.\n\n" "This operation will override any previous " "installation choices about which drives to " "ignore.\n\n" "Would you like to initialize this drive, " - "erasing ALL DATA?") % (drive,) + "erasing ALL DATA?") % (drive, dev.model, getDeviceSizeMB (dev),) - rc = 0 - if (ks and (drive in clearDevs) and initAll) or \ - self.isDisciplineFBA(drive): - rc = 1 - else: - if not intf: - self._removeDisk(drive) - return False - - deviceFile = "/dev/" + drive - dev = parted.PedDevice.get(deviceFile) - - msg = _("The partition table on device %s (%s %-0.f MB) was unreadable.\n" - "To create new partitions it must be initialized, " - "causing the loss of ALL DATA on this drive.\n\n" - "This operation will override any previous " - "installation choices about which drives to " - "ignore.\n\n" - "Would you like to initialize this drive, " - "erasing ALL DATA?") % (drive, dev.model, getDeviceSizeMB (dev),) + rc = intf.messageWindow(_("Warning"), msg, type="yesno") if rc != 0: return True -- 1.5.4.3

2 1

PATCH: only allow selection of devices known by booty as bootloader target disk
by Hans de Goede 12 Sep '08

12 Sep '08

Hi All, As I do not have any commit access to git as of yet, I'm sending my patches to the list for now. The attached patch only allows selection of devices known by booty as bootloader target disk, fixing the crash reported in bug 454135 Regards, Hans diff --git a/iw/autopart_type.py b/iw/autopart_type.py index c1366f7..714d04a 100644 --- a/iw/autopart_type.py +++ b/iw/autopart_type.py @@ -365,6 +365,8 @@ class PartitionTypeWindow(InstallWindow): else: defaultBoot = None for disk in self.diskset.disks.values(): + if not disk.dev.path[5:] in self.anaconda.id.bootloader.drivelist: + continue size = partedUtils.getDeviceSizeMB(disk.dev) dispstr = "%s %8.0f MB %s" %(disk.dev.path[5:], size, disk.dev.model) i = bootstore.append(None)

2 2

RE: buildinstall/mk-images and isolinux.cfg problems
by Moray Henderson (ICT) 11 Sep '08

11 Sep '08

>From: treed(a)ultraviolet.org >>>Building initrd.img >>>Module 3c501 not found in kernel rpm >>>Module 3c503 not found in kernel rpm >>>Module 3c505 not found in kernel rpm >> Is your build system the same kernel and architecture as the release >you're trying to build? > >Yes. 2.6.18-92.1.6.el5 x86_64 for both. Are those 3c501, 3c503, 3c505 and the other modules supposed to be part of the standard kernel, or did you get them from somewhere else? Hmm, you'll probably get a better answer from someone who actually knows the relationship between modules in the running kernel and modules in initrd.img on x86_64 architecture. >Ah-hah...Well that seems rather silly. Maybe I shouldn't be using >buildinstall at all for what I'm trying to do (customize the CD)? If you're doing it this way, you're supposed to need buildinstall to provide the images your CD needs to boot. There is at least one hidden file, /.discinfo, that needs to be present and usually gets missed if you just copy the contents of a CD. I don't know what other manual steps would be necessary. There is a tool called Revisor (http://revisor.fedoraunity.org/) which may help you if you are running a graphical environment. I'm told it can handle CentOS. > To perform a CD-ROM-based kickstart installation, the kickstart file >must > be named ks.cfg and must be located in the boot CD-ROM's top-level > directory. You're right, that does sound weird. I think I see the problem, though: you need an extra slash in the URI for your kickstart file. At least, what works for me is a kickstart file /isolinux/install.cfg referred to using ks=cdrom://isolinux/install.cfg. I know, that's not what it says on http://www.centos.org/docs/5/html/Installation_Guide-en-US/s1-kickstart2-st…. Moray. "To err is human. To purr, feline" http://members.aol.com/edgwddirk

2 1

[PATCH] Support for more than 8 cciss (SmartArray) devices
by Radek Vykydal 11 Sep '08

11 Sep '08

Hi, major numbers for more than 8 cciss devices are assigned dynamically, thus read from /proc/devices. Static table is used to cache values for up to 32 cciss devices (the number recent driver supports). The patch builds, but was not tested. Radek

1 0

RE: buildinstall/mk-images and isolinux.cfg problems
by Moray Henderson (ICT) 11 Sep '08

11 Sep '08

>From: treed(a)ultraviolet.org > >unpacking >/var/www/html/centos/5.2/os/x86_64/CentOS/kernel-2.6.18- >92.el5.x86_64.rpm.x86_64 >Building initrd.img >Module 3c501 not found in kernel rpm >Module 3c503 not found in kernel rpm >Module 3c505 not found in kernel rpm Is your build system the same kernel and architecture as the release you're trying to build? >But after buildinstall runs the isolinux.cfg is changed back! > >Where is it getting the original from? Where should I be changing it to >get my modified isolinux.cfg into my cd? There's a syslinux.cfg (along with other files that buildinstall overwrites) in /usr/lib/anaconda-runtime/boot. I keep my own copies of those files, and copy them back in to /isolinux when buildinstall has finished. >Just to get around the isolinux.cfg issue manually I tried booting and at >the prompt I say: > >linux ks=cdrom:/kickstart.cfg > >But it says that it cannot find the kickstart.cfg file. That looks as if it ought to work for a kickstart.cfg file in the root of the CD. Problem with the file itself perhaps? Moray. "To err is human. To purr, feline" http://members.aol.com/edgwddirk

2 2

CD check question
by Ali Majdzadeh 10 Sep '08

10 Sep '08

Hello All Is it possible to by pass the CD check phase? Kind Regards Ali Majdzadeh Kohbanani

1 1

How does anaconda decide to upgrade vs install? (Cannot upgrade F7 to F9)
by Martin Langhoff 10 Sep '08

10 Sep '08

Using Revisor, I have built an custom install CD that intend will be able to upgrade a custom F7 system. Anaconda, is missing the cue and wants to perform a fresh install... The custom Fedora spins are - The current version of the OLPC school server - xs-0.4, based on F7. Originally installed from a livecd. - The development version of the OLPC school server - a F9-based spin. The existing disk has sda1 /boot sda2 / ... and in /etc/redhat-release it contains the 'Fedora 7 (Moonshine)' string In the anaconda sources, I am looking at dispatch.py, backend.py and upgrade.py but it's not 100% clear to me what the rules are. cheers, m -- martin.langhoff(a)gmail.com martin(a)laptop.org -- School Server Architect - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff

2 4

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Anaconda-devel September 2008