https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c1 Jiaying ren changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO CC| |jren@suse.com InfoProvider| |hw01@heisch.inka.de --- Comment #1 from Jiaying ren 2012-12-26 08:18:37 UTC --- (In reply to comment #0)

User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; de; rv:1.9.2.13) Gecko/20101203 SUSE/3.6.13-0.1.1 Firefox/3.6.13

Starting Situation: ------------------- /dev/sda contains a (productive) opensuse 11.1 (Kde 3.5 :-) /dev/sdb has had a working opensuse 12.1 (only for software tests KDE4.. :-(

Boot loader went to /dev/sdb, and from there to /dev/sda to start 11.1.

Hi~Werner. Thank you for your report. Would you please attach your grub/grub2 config? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c Jiaying ren changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|hw01@heisch.inka.de | AssignedTo|bnc-team-screening@forge.pr |mchang@suse.com |ovo.novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c4 --- Comment #4 from Michael Chang 2012-12-27 11:09:40 UTC --- Werner, It could be udev related and please help to get output of "udisksctl dump" on your failure 12.2 system ? $ udisksctl dump > udisk.dmp I suspect that udev change would cause the 12.1 device.map fail to look up corresponding kernel device under 12.2. (Thus hd0 is used instead as fallback ..) Of course, logs under /var/log/YaST2 are appreciated .. Thanks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c6 --- Comment #6 from Werner Heisch 2013-01-01 01:20:13 UTC --- Hi Michael, (with the best wishes for the new year), I have attached the requested files in a tgz-archive. ( I have no clue about such things, but ) for me, "udisk.dmp" looks OK More interesting I find ../Yast2/perl-BL-standalone.log ------------------------------ What you see, is the second try on /dev/sdb = ST31000 ... - installation of opensuse 12.1 and updating to 12.2 All worked fine for 12.1. The first error I recognized is in line 29927. Problem : (hd0,0) should be (hd1,0) ( As mentioned before,it can be fixed by editing the resulting /boot/grub/menu.lst.) BTW: I believe, it would be a good idea, to decide between /dev/disk/by-id/ ... and /dev/sd(a.b.c....) while istalling the system. Reason: Specially in small companies, there are a lot of stand-alone computers, i.e. for error reporting, production planning, visualization which are not supported by y big IT team. It is a good practice, to swap the hard disks against new ones, after some time. This is a prophylactic service. This costs small money ( = the hd) but saves big money, that will be lost if the production is not possible because hard disk is damaged. In the good old times, a hd was cloned by "dd if=/dev/(src) of=/dev(dest)", today it's a big thing, if you are not doing this job every day. Using /dev/disk/by-id/.. probably will be right, if you are running a main frame. I assume, that 99% of the linux user do not have one, but are bothered by this (for the useless) complications. We are on the way that linux is becoming more an more complicated and it's not only me, who thinks that this is not the right way. Some of my friends ( experienced soft workers, over 30years experience ) frist worked with HP-UX, later with Linux, now with windows, because the felt messed around by continuous changes and featuritis. One them sayed: Linux is not Unix. Unix rule : "Never touch a running system" Linux rule : "If it is stable, change it" Greetings Werner -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c8 Michael Chang changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |snwint@suse.com --- Comment #8 from Michael Chang 2013-01-02 11:09:35 UTC --- Wener,

From the very beginning the log has pointed to the root cause ..

2012-12-24 16:37:31 MILESTONE: *** WARNING: No UDEV mapping! *** It cause perl-Bootlader's GetKernelDevice() failed to translate device names to kernel device names 2012-12-24 16:53:58 MILESTONE: GRUB::GetKernelDevice: From /dev/disk/by-id/ata-ST31000528AS_5VP2Y163 to /dev/disk/by-id/ata-ST31000528AS_5VP2Y163 It should translate to sdb and output sth like this GRUB::GetKernelDevice: From /dev/disk/by-id/ata-ST31000528AS_5VP2Y163 to /dev/sdb Finally the fallback device is used (hd0) as no matched "kernel device" can be found in udevmap 2012-12-24 16:53:58 WARNING: GRUB::UnixDev2GrubDev: Unknown device/partition, using fallback 2012-12-24 16:53:58 MILESTONE: GRUB::UnixDev2GrubDev: Translated UNIX partition -> GRUB device: /dev/disk/by-id/ata-ST31000528AS_5VP2Y163-part1 to (hd0,0) 2012-12-24 16:53:58 MILESTONE: Set GRUB's root to (hd0,0) Probably udevadm info -q all -n $dev 2>/dev/null Has no output *during the update*, I can't tell why .. And an attempt to fix the issue in git commit from Steffen https://github.com/openSUSE/perl-bootloader/commit/46c82d61f4a9e12999aec16f2... Steffen any other good idea? I conclude this is udev problem for now. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

...

...

...

...

...

>>>>>>>>>>>>>>>>>>>> /dev/sda2 swap swap defaults 0 0 /dev/sda1 / ext3 acl,user_xattr 1 1 /dev/sda3 /home ext3 acl,user_xattr 1 2 /dev/sda5 /v1 ext3 acl,user_xattr 1 2 /dev/sda6 /v2 ext3 acl,user_xattr 1 2

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c10 Werner Heisch changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|snwint@suse.com | --- Comment #10 from Werner Heisch 2013-01-05 16:10:34 UTC --- Hi Michael, Hi Steffen, maybe an reason: Before installing 12.2 I have cloned my productive system on /dev/sda to a new disk. I've edited /etc/fstab not using /dev/disk/by-id/.. because I do not like it. My (sda1) /etc/fstab looks like this: proc /proc proc defaults 0 0 sysfs /sys sysfs noauto 0 0 debugfs /sys/kernel/debug debugfs noauto 0 0 usbfs /proc/bus/usb usbfs noauto 0 0 devpts /dev/pts devpts mode=0620,gid=5 0 0 /dev/disk/by-id/ata-SanDisk_SDSSDP064G_121977404574-part2 /ssd ext3 acl,user_xattr 1 2 <<<<<<<<<<<<<<<<<<<<<<<<<<< My (sda1) boot/grub/device.map looks like this:

...

...

...

...

...

>>>>>>>>>>>>>>>>>>>> (hd2) /dev/disk/by-id/ata-SanDisk_SDSSDP064G_121977404574 (hd1) /dev/disk/by-id/ata-ST31000528AS_5VP2Y163 (hd0) /dev/disk/by-id/ata-ST500DM002-1BD142_W2AFJ83S <<<<<<<<<<<<<<<<<<<<<<<<<<<

My (sda1) boot/grub/menu.lst looks like this:

...

...

...

...

...

>>>>>>>>>>>>>>>>>>>> # Modified by YaST2. Last modification on Fr Dez 14 04:02:07 CET 2012 default 0 timeout 8 gfxmenu (hd0,0)/boot/message

###Don't change this comment - YaST2 identifier: Original name: linux### title SUSE LINUX 11.1 root (hd0,0) kernel /boot/vmlinuz root=/dev/disk/by-id/ata-ST500DM002-1BD142_W2AFJ83S-part1 repair=1 resume=/dev/disk/by-id/ata-ST500DM002-1BD142_W2AFJ83S-part2 splash=silent showopts vga=0x31a initrd /boot/initrd ###Don't change this comment - YaST2 identifier: Original name: failsafe### title Failsafe -- SUSE LINUX root (hd0,0) kernel /boot/vmlinuz root=/dev/disk/by-id/ata-ST500DM002-1BD142_W2AFJ83S-part1 showopts ide=nodma apm=off noresume edd=off powersaved=off nohz=off highres=off processor.max_cstate=1 x11failsafe vga=0x31a initrd /boot/initrd title opensuse 12.2 root (hd0,0) kernel /boot/vmlinuz root=/dev/sdb1 repair=1 resume=/dev/sdb1 splash=silent showopts vga=0x31a initrd /boot/initrd <<<<<<<<<<<<<<<<<<<<<<<<<<< -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c12 Michael Chang changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |fcrozat@suse.com --- Comment #12 from Michael Chang 2013-01-07 09:11:25 UTC --- Werner, It's really hard to figure out why udevmap is not available per your log at that moment. Without that info perl bootloader lamed by not able to translate by-id (and by-label, by-uuid) names to kernel device name. I believed the problem has been "avoided" in 12.2 and later as per Steffen's commit, it's not counting on udev provided info but directly transverse the /dev tree. This may not be generally bootloader's business but a system-wide problem (related with udev, package update process and so on ..) CC udev maintainer here, Hi Frederic, Did we ever come across similar problem, more specifically, this command udevadm info -q all -n $dev 2>/dev/null has no output during update (12.1 => 12.2) ? THANKs. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c13 Frederic Crozat changed: What |Removed |Added ---------------------------------------------------------------------------- InfoProvider|fcrozat@suse.com |rmilasan@suse.com --- Comment #13 from Frederic Crozat 2013-01-07 14:27:43 UTC --- reassigning to robert, I'm not maintaining udev -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c15 --- Comment #15 from Michael Chang 2013-01-08 05:07:36 UTC --- Robert, Sorry for incomplete message, it's GetUdevMapping() function in perl Bootloader (Tools.pm). I depict some relevant code below .. ==== my %mapping = (); my @output = `find -P /dev -type b`; .. for my $dev (@output) { .. my @output2 = `udevadm info -q all -n $dev 2>/dev/null` .. for (@output2) { $mapping{"/dev/$1"} = $dev if /S:\s(.*)$/ } } .. if (!(keys %mapping)) { logger->milestone("*** WARNING: No UDEV mapping! ***"); .. } === I really don't know what should be blamed here, as it's not likely to fail anyhow. The one possibility to me is that udev returns no useful info. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c17 --- Comment #17 from Michael Chang 2013-01-08 06:47:36 UTC --- Robert, Hm. It happens to me that this is very much likely a dup of bnc#591700. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=795838 https://bugzilla.novell.com/show_bug.cgi?id=795838#c19 --- Comment #19 from Robert Milasan 2013-01-08 08:06:41 UTC --- I think this has been fix by Stefan, we had a similar issue before, I just don't remember the bug number. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.