[opensuse] Very strange boot roblem
Hello SuSE people, I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system) Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic. That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software? When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please? Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/07 23:45 (GMT-0500) Bob S composed:
I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system)
Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic.
Boot 11.0 and find out the name of the partition on which 11.2 is installed, e.g. /dev/sda6. Then boot to the 11.2 Grub menu, and substitute that at the root= position. Once you have 11.2 booted you can repair the Grub menu and fstab if necessary.
That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software?
You're hitting fallout from widespread changes made to support booting from USB and other media that may or may not be present on any given boot. Apparently part of the disk label for that partition in /dev is tied to the controller on the dead motherboard, but its the label hiding in the initrd, not the applicable one, if I'm right about what's going on. It may mean you won't be able to boot using the old initrd and will have to rescue boot to build a new one, depending on whether the initrd supports the necessary device name.
When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please?
Exactly when was the old motherboard made? Make? Model? If it's several years old it could be victim of premature cap failure, nothing to do with actually using or misusing it. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tuesday 08 December 2009 01:04:45 Felix Miata wrote:
On 2009/12/07 23:45 (GMT-0500) Bob S composed:
I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system)
Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic.
Boot 11.0 and find out the name of the partition on which 11.2 is installed, e.g. /dev/sda6. Then boot to the 11.2 Grub menu, and substitute that at the root= position. Once you have 11.2 booted you can repair the Grub menu and fstab if necessary.
Hi Felix, Thanks for replying. Seems you always come to my rescue. The boot partition is indelibly pasted in my mind. (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ? FYI every partition on every disk is mounted no matter what OS is mounted and run. I can see everything and access everything. I can look at the 11.2 /boot/grub/menu.lst or the fstab or whatever. I have made sure that every menu.lst is exactly the same for every OS.
That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software?
You're hitting fallout from widespread changes made to support booting from USB and other media that may or may not be present on any given boot. Apparently part of the disk label for that partition in /dev is tied to the controller on the dead motherboard, but its the label hiding in the initrd, not the applicable one, if I'm right about what's going on. It may mean you won't be able to boot using the old initrd and will have to rescue boot to build a new one, depending on whether the initrd supports the necessary device name.
When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please?
Exactly when was the old motherboard made? Make? Model? If it's several years old it could be victim of premature cap failure, nothing to do with actually using or misusing it. -- The old motherboard was an MSI K8T with a 939 pin AMD processor, about 4 years
OK, I have noticed that the new boot sequence is different from what I am used to seeing. Lots of USB stuff which is kind of confusing to me. So, can I build a new initrd for 11.2 ? You spoke of rescue boot for grub. Is that part of the grub command? Guess I will have to review my grub options. old. Can't believe that the 939 is "passe" Was definitely caps. The new board is also MSI "V" series with a dual core AMD processor. Any further instruction/guidance is certainly appreciated. Bob S PS I really like and enjoy your sigs. Keep them coming. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
.... (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ?
The kernel lines in your 11.2 menu.lst stanzas include something like: kernel...vmlinuz...root=/dev/disk/yadayadayadayadayadayadayadayadayada showopts... While in the Grub menu part of the boot process, substitute /dev/sdc8 for the .../yadas in your menu choice and report back what happens when you proceed. Oh, wait a minute. SUSE stupidly puts the showopts parameter after the root= on kernel lines, and moves it back every time perl-Bootloader gets run after manually putting it in the more useful pre-root= position. That prevents you from seeing the root= section while in the GFX Grub menu. To do what I wrote requires you exit the GFX Grub menu to the text Grub menu to edit the kernel line root= section. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wednesday 09 December 2009 03:03:12 Felix Miata wrote:
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
.... (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ?
The kernel lines in your 11.2 menu.lst stanzas include something like:
kernel...vmlinuz...root=/dev/disk/yadayadayadayadayadayadayadayadayada showopts...
While in the Grub menu part of the boot process, substitute /dev/sdc8 for the .../yadas in your menu choice and report back what happens when you proceed.
Oh, wait a minute. SUSE stupidly puts the showopts parameter after the root= on kernel lines, and moves it back every time perl-Bootloader gets run after manually putting it in the more useful pre-root= position. That prevents you from seeing the root= section while in the GFX Grub menu.
To do what I wrote requires you exit the GFX Grub menu to the text Grub menu to edit the kernel line root= section. -- I have no idea what you are telling me here, not understanding the "SUSE way" and do not know what you mean by the GFX Grub menu. If you mean editing
Felix I edited the menu.lst as you suggested. Changing /dev/disk/yadayada to /dev/sdc8. Tried to boot and now it tells me it cannot find /dev/sdc8 rather than the /dev/disk/by-label it did previously. Interestingly the line above it in the boot sequence also says it cannot find /dev/sdc3 (which is the resume line) in menu.lst) and states ignoring then continues on to the next line about not finding /dev/sdc8. menu.lst that is easy. I have tried every iteration of configuration I can possibly think of. Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/10 22:43 (GMT-0500) Bob S composed:
Felix Miata wrote:
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
.... (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ?
The kernel lines in your 11.2 menu.lst stanzas include something like:
kernel...vmlinuz...root=/dev/disk/yadayadayadayadayadayadayadayadayada showopts...
While in the Grub menu part of the boot process, substitute /dev/sdc8 for the .../yadas in your menu choice and report back what happens when you proceed.
I edited the menu.lst as you suggested.
I'd like you to point me to the archive URL of the post where you think I suggested editing menu.lst. I don't remember any such thing. If you are going to edit menu.lst, move all the showopts in front of root= so that in the future you can easily edit root=yadayada... while using the menu to boot.
Changing /dev/disk/yadayada to /dev/sdc8. Tried to boot and now it tells me it cannot find /dev/sdc8 rather than the /dev/disk/by-label it did previously.
OK. Next, do what I actually suggested.
Interestingly the line above it in the boot sequence also says it cannot find /dev/sdc3 (which is the resume line) in menu.lst) and states ignoring then continues on to the next line about not finding /dev/sdc8.
Add noresume to your kernel lines in grub and you'll get fewer messages.
Oh, wait a minute. SUSE stupidly puts the showopts parameter after the root= on kernel lines, and moves it back every time perl-Bootloader gets run after manually putting it in the more useful pre-root= position. That prevents you from seeing the root= section while in the GFX Grub menu.
To do what I wrote requires you exit the GFX Grub menu to the text Grub menu to edit the kernel line root= section.
I have no idea what you are telling me here, not understanding the "SUSE way" and do not know what you mean by the GFX Grub menu.
GFX Grub menu is the screen from which you choose a Grub stanza to boot, or whatever else is in the Grub menu, like chainloading Windows or running Memtest. It is distinguished from standard grub menus in that standard grub menus have no more than four colors, commonly only two colors, no Function key submenu across the bottom of the screen, and no edit line for the currently selected kernel.
I have tried every iteration of configuration I can possibly think of.
This is where understanding how to use the Grub prompt is helpful, or more. Exit the Grub GFX menu, and exit the Grub text menu to the grub prompt. When you get there, do: find /boot/grub/stage1 That will list locations where Grub has been "installed". If you know what to do next, good. If not, just CAD to reboot. Then either try to figure out which of those is the actual location of your 11.2 partition, and substitute both the root= device name on the kernel line and the root (x,y) line that corresponds to it on your next attempt(s) to boot, in the manner I indicated previously (on the fly) as opposed to actually editing menu.lst, or report back here what it lists for further instructions. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday 11 December 2009 00:28:33 Felix Miata wrote:
On 2009/12/10 22:43 (GMT-0500) Bob S composed:
Felix Miata wrote:
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
.... (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ?
The kernel lines in your 11.2 menu.lst stanzas include something like:
kernel...vmlinuz...root=/dev/disk/yadayadayadayadayadayadayadayadayada showopts...
While in the Grub menu part of the boot process, substitute /dev/sdc8 for the .../yadas in your menu choice and report back what happens when you proceed.
I edited the menu.lst as you suggested.
I'd like you to point me to the archive URL of the post where you think I suggested editing menu.lst. I don't remember any such thing.
OK. guess I misunderstood you. From a few lines up is"> >> The kernel lines in your 11.2 menu.lst stanzas include something like:
If you are going to edit menu.lst, move all the showopts in front of root= so that in the future you can easily edit root=yadayada... while using the menu to boot.
OK, that is not a big deal.
Changing /dev/disk/yadayada to /dev/sdc8. Tried to boot and now it tells me it cannot find /dev/sdc8 rather than the /dev/disk/by-label it did previously.
So that is why I changed it in menu.lst
OK. Next, do what I actually suggested.
Hmmm.l...confusion reigns....Not sure what you actually suggested.
Interestingly the line above it in the boot sequence also says it cannot find /dev/sdc3 (which is the resume line) in menu.lst) and states ignoring then continues on to the next line about not finding /dev/sdc8.
Add noresume to your kernel lines in grub and you'll get fewer messages.
OK. But not a big deal. Don't really care. Just told you for info. /dev/sdc3 is actually my swap partition.
Oh, wait a minute. SUSE stupidly puts the showopts parameter after the root= on kernel lines, and moves it back every time perl-Bootloader gets run after manually putting it in the more useful pre-root= position. That prevents you from seeing the root= section while in the GFX Grub menu.
To do what I wrote requires you exit the GFX Grub menu to the text Grub menu to edit the kernel line root= section.
I have no idea what you are telling me here, not understanding the "SUSE way" and do not know what you mean by the GFX Grub menu.
GFX Grub menu is the screen from which you choose a Grub stanza to boot, or whatever else is in the Grub menu, like chainloading Windows or running Memtest. It is distinguished from standard grub menus in that standard grub menus have no more than four colors, commonly only two colors, no Function key submenu across the bottom of the screen, and no edit line for the currently selected kernel.
I have tried every iteration of configuration I can possibly think of.
This is where understanding how to use the Grub prompt is helpful, or more.
Sure wish I could. Have read all I could find, including stuff you have posted previouly to other folks.
Exit the Grub GFX menu, and exit the Grub text menu to the grub prompt. When you get there, do:
find /boot/grub/stage1
That will list locations where Grub has been "installed". If you know what to do next, good. If not, just CAD to reboot. Then either try to figure out which of those is the actual location of your 11.2 partition, and substitute both the root= device name on the kernel line and the root (x,y) line that corresponds to it on your next attempt(s) to boot, in the manner I indicated previously (on the fly) as opposed to actually editing menu.lst, or report back here what it lists for further instructions. --
OK, when I did that, this is what I got: ------------------------------------------------------------ grub> find /boot/grub/stage1 (hd1,0) (hd2,7) grub> ---------------------------------------------------- Which is correct. (hd2,7) is the 11.2 partition. Don't know where to go from here. Bob S Really hope I am not being stupid....stupid....alzhiemers setting in here. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/10 22:24 (GMT-0500) Bob S composed:
Felix Miata wrote:
Exit the Grub GFX menu, and exit the Grub text menu to the grub prompt. When you get there, do:
find /boot/grub/stage1
That will list locations where Grub has been "installed". If you know what to do next, good. If not, just CAD to reboot. Then either try to figure out which of those is the actual location of your 11.2 partition, and substitute both the root= device name on the kernel line and the root (x,y) line that corresponds to it on your next attempt(s) to boot, in the manner I indicated previously (on the fly) as opposed to actually editing menu.lst, or report back here what it lists for further instructions.
OK, when I did that, this is what I got: ------------------------------------------------------------ grub> find /boot/grub/stage1 (hd1,0) (hd2,7)
grub> ---------------------------------------------------- Which is correct. (hd2,7) is the 11.2 partition. Don't know where to go from here.
Since root=/dev/sdc8 fails with "cannot find", try root=/dev/sdb8 instead to see what happens. You do have only two attached HDs, right? What do you have connected to USB during boot, an unpopulated memory chip reader maybe? If you have only two disks, there should be no (hd2), only (hd0) and (hd1), the latter of which would have your ,7 11.2 / partition. This kind of thing is precisely why the default configuration switchover to using /dev/disk* for root= and fstab entries instead of device names. Where are your HDs attached? PATA ports hda & hdc? PATA ports hda & hdb? 2 SATA ports? PATA hda & SATA both? If the latter, which OS's HD is on which port? Using both PATA & SATA HDs in the same system can be monumental multiboot pain. I wonder if when you boot the DVD to install or repair that the DVD steals (hd0)/sda, leaving the 2 HDs with (hd1)/sdb & (hd2)/sdc instead of (hd0)/sda & (hd1)/sdb as they would be on a normal boot? -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday 11 December 2009 07:29:00 Felix Miata wrote:
On 2009/12/10 22:24 (GMT-0500) Bob S composed:
Felix Miata wrote:
Exit the Grub GFX menu, and exit the Grub text menu to the grub prompt. When you get there, do:
find /boot/grub/stage1
That will list locations where Grub has been "installed". If you know what to do next, good. If not, just CAD to reboot. Then either try to figure out which of those is the actual location of your 11.2 partition, and substitute both the root= device name on the kernel line and the root (x,y) line that corresponds to it on your next attempt(s) to boot, in the manner I indicated previously (on the fly) as opposed to actually editing menu.lst, or report back here what it lists for further instructions.
OK, when I did that, this is what I got: ------------------------------------------------------------ grub> find /boot/grub/stage1 (hd1,0) (hd2,7)
grub> ---------------------------------------------------- Which is correct. (hd2,7) is the 11.2 partition. Don't know where to go from here.
Since root=/dev/sdc8 fails with "cannot find", try root=/dev/sdb8 instead to see what happens.
Tried sda8, sdb8, sdc8,and disk/by-label/11.2 All with the same result.
You do have only two attached HDs, right?
No, I have 3. Two ide (80GB & 150GB) designated sda and sdb in that order, and one sata (250GB) designated sdc. The 11.2 is ond sdc8.
What do you have connected to USB during boot, an unpopulated memory chip reader maybe?
Yes an unpopulated USB card reader
If you have only two disks, there should be no (hd2), only (hd0) and (hd1), the latter of which would have your ,7 11.2 / partition. This kind of thing is precisely why the default configuration switchover to using /dev/disk* for root= and fstab entries instead of device names.
Yes, every Linux partition on all drives is designated /dev/disk/by-label and they are all reflected that way in fstab.
Where are your HDs attached? PATA ports hda & hdc? PATA ports hda & hdb? 2 SATA ports? PATA hda & SATA both? If the latter, which OS's HD is on which port? Using both PATA & SATA HDs in the same system can be monumental multiboot pain.
I wonder if when you boot the DVD to install or repair that the DVD steals (hd0)/sda, leaving the 2 HDs with (hd1)/sdb & (hd2)/sdc instead of (hd0)/sda & (hd1)/sdb as they would be on a normal boot? -- Well, something happens.And I can't explain it. If I go into the rescue system and choose "boot installed system" for 11.2 it is shown as sda8. It boots 11.2 perfectly. Why it shows sda8 instead of sdc8 I have no idea. Very illogical. Remember now, I tried to boot normally as sda8 and got that stupid message about finding it. The partitioner, fdisk, and Parted Magic all show
The two PATA ports are hda, hdb (sda, sdb) There are only two on this new board. The rest are all SATA. (Had to buy a new SATA DVD writer) the disk layout correctly. This is crazy. Why should replacing the MB have anything to do with booting? That is all HD stuff except for the bios. And that all seems to be correct. And why does the install/rescue system mis-identify tue disks in a seemingly random order? Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/12 02:44 (GMT-0500) Bob S composed:
Felix Miata wrote:
Since root=/dev/sdc8 fails with "cannot find", try root=/dev/sdb8 instead to see what happens.
Tried sda8, sdb8, sdc8,and disk/by-label/11.2 All with the same result.
Because of the USB chip reader, maybe you should try /dev/sdd8 too. Or try with the reader unplugged, or both.
You do have only two attached HDs, right?
No, I have 3.
Why Grub found only two locations for /boot/grub/stage1, on hd1 & hd2, it made me think you had only two and wonder why no hd0. You have Windoz on a separate disk? Why no /boot/grub/stage1 anywhere on hd0?
Two ide (80GB & 150GB) designated sda and sdb in that order, and
Under 11.0 and 11.2, or only under 11.2, or only under 11.0?
one sata (250GB) designated sdc. The 11.2 is ond sdc8.
This, plus the unpopulated USB card reader, is exactly why modern installers use /dev/disk/by-* instead of device names.
What do you have connected to USB during boot, an unpopulated memory chip reader maybe?
Yes an unpopulated USB card reader
Does 11.0 boot OK if you have a chip in it when you boot? What is it's device name in 11.0. Can you access it during a 11.2 rescue boot? If so, what device name does it get?
If you have only two disks, there should be no (hd2), only (hd0) and (hd1), the latter of which would have your ,7 11.2 / partition. This kind of thing is precisely why the default configuration switchover to using /dev/disk* for root= and fstab entries instead of device names.
Yes, every Linux partition on all drives is designated /dev/disk/by-label and they are all reflected that way in fstab.
I don't recall anyone else reporting failure using /dev/disk/by-label before this thread. :-(
Where are your HDs attached? PATA ports hda & hdc? PATA ports hda & hdb? 2 SATA ports? PATA hda & SATA both? If the latter, which OS's HD is on which port? Using both PATA & SATA HDs in the same system can be monumental multiboot pain.
The two PATA ports are hda, hdb (sda, sdb) There are only two on this new board. The rest are all SATA. (Had to buy a new SATA DVD writer)
Does it make any difference if you swap ports between the DVD & the SATA HD?
I wonder if when you boot the DVD to install or repair that the DVD steals (hd0)/sda, leaving the 2 HDs with (hd1)/sdb & (hd2)/sdc instead of (hd0)/sda & (hd1)/sdb as they would be on a normal boot?
Well, something happens.And I can't explain it. If I go into the rescue system and choose "boot installed system" for 11.2 it is shown as sda8. It boots 11.2 perfectly. Why it shows sda8 instead of sdc8 I have no idea. Very
It's not using an initrd made from your old motherboard. The generic ones that come with installation programs seem to be quite a bit smarter. Maybe the DVD is another fly in the ointment, taking or not a logical position opposite that when booting with no DVD in the drive. What happens when you try to boot from 11.2 by first booting from DVD and then choosing boot from HD from the DVD boot menu?
illogical. Remember now, I tried to boot normally as sda8 and got that stupid message about finding it. The partitioner, fdisk, and Parted Magic all show the disk layout correctly.
How did you decide what is "correct"? Normally in mixed PATA/SATA environments the BIOS will default to SATA first, aka your SATA HD would be sda or sdb, your sr0 DVD would be sda or sdb, then your PATA would be sdc & sdd, but all only if no USB intervention. With the USB reader you're probably screwing it royally as long as not using by-uuid or by-id, though by-label ought to work.
This is crazy. Why should replacing the MB have anything to do with booting? That is all HD stuff except for the bios. And that all seems to be correct. And why does the install/rescue system mis-identify tue disks in a seemingly random order?
Most installation programs now default to not using either device names or by-label, but instead by-id or by-uuid. I suggest you try editing menu.lst and substituting by-uuid for whatever now follows root= and if that fails, try by-id, and if that too fails, the last resort, by-path. When you installed 11.2, were all 3 HDs connected? Maybe a repair install from the DVD would work if you first disconnect the two PATA drives and the USB reader, which would force the SATA to be sda. Then after that's done, you confirm it boots itself with nothing else connected, then reconnect, if it doesn't boot successfully as sda, then go into the BIOS and explicitly set the order to be SATA first. Can you try the other options for ATA controller configuration? IOW, if what is set now is called "enhanced mode", try whatever else is there to choose from. If now it's set to "compatible", try the other. If now it's set to "AHCI", try the other. If you can't get it to work using /dev/disk/by-uuid*, I suspect it might be easier to reinstall (with no connected PATAs, no connected USB reader, and the HD on the first SATA port) than figure it out. I wish it was here to get my hands on it. You're not by any chance in central FL somewhere are you? If any of the above sounds like rambling or nonsense, it's because I should have been in bed at least 2 hours ago, and where I'm off to now. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Sat, 12 Dec 2009 02:44:56 -0500, you wrote:
This is crazy. Why should replacing the MB have anything to do with booting?
One reason could be because devices are initialized in a diffeent order. Another one could be because driver initialization changed.
That is all HD stuff except for the bios. And that all seems to be correct. And why does the install/rescue system mis-identify tue disks in a seemingly random order?
I can't answer that, but I *urge* you to use the symlinks in /dev/disk/by-id as these are generated from the attached devices and thus stay the same. Philipp -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Friday 11 December 2009 00:28:33 Felix Miata wrote:
On 2009/12/10 22:43 (GMT-0500) Bob S composed:
Forgot to add. Hereis the stanza from menu.lst. ------------------------------------------------------------------------ ###Don't change this comment - YaST2 identifier: Original name: linux### title openSUSE 11.2 root (hd2,7) kernel /boot/vmlinuz-2.6.31.5-0.1-desktop root=/dev/disk/by-label/11.2 resume=/dev/sdc3 splash=silent showopts vga =0x314 vga=0x314 initrd /boot/initrd-2.6.31.5-0.1-desktop ----------------------------------------------------------------------------------------- Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/10 22:30 (GMT-0500) Bob S composed:
Forgot to add. Hereis the stanza from menu.lst. ------------------------------------------------------------------------ ###Don't change this comment - YaST2 identifier: Original name: linux### title openSUSE 11.2 root (hd2,7) kernel /boot/vmlinuz-2.6.31.5-0.1-desktop root=/dev/disk/by-label/11.2 resume=/dev/sdc3 splash=silent showopts vga =0x314 vga=0x314 initrd /boot/initrd-2.6.31.5-0.1-desktop -----------------------------------------------------------------------------------------
At least until you've gotten through the current mess: 1-move showopts ahead of root= 2-replace resume=/dev/sdc3 with noresume 3-remove splash=silent 4-remove vga =0x314, leaving vga=0x314 Recommendation: In the future when setting partition labels, don't use punctuation marks, do use at least some letters, and don't be overly cryptic. e.g. instead of 11.2, do 112root or p08suse112 -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Friday, 2009-12-11 at 07:28 -0500, Felix Miata wrote:
At least until you've gotten through the current mess: 1-move showopts ahead of root= 2-replace resume=/dev/sdc3 with noresume
Why? That breaks hibernation, which I find very useful. What I would do is change it to by-label. And anyway, it does not have any effect on booting.
3-remove splash=silent
Change to splash=verbose. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkskA0oACgkQtTMYHG2NR9UfDQCfYK2xa2995SAr+rgZOJr+Jw5V cnAAnjMtNw7FQuwqs2rXssPiueVUHMsE =ZW1A -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/12 21:55 (GMT+0100) Carlos E. R. composed:
On Friday, 2009-12-11 at 07:28 -0500, Felix Miata wrote:
At least until you've gotten through the current mess: 1-move showopts ahead of root= 2-replace resume=/dev/sdc3 with noresume
Why? That breaks hibernation, which I find very useful. What I would do is change it to by-label. And anyway, it does not have any effect on booting.
Hibernation doesn't work when you can't even boot. Here in this thread we're just trying to get to a booted state and eliminating clutter during the process of getting there could make a helpful difference.
3-remove splash=silent
Change to splash=verbose.
I think if you look that up you'll find it doesn't actually exist, and that it works is because it matches the default, which is nothing at all. Still, splash=0 and splash=off have the same effect, while leaving the too long kernel lines with strings like root=/dev/disk/by-label/ata-ST3320620AS_6QF3PJXC-part10 in menu.lst shorter, maybe shorter enough to fit on one line instead of wrapping. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Saturday, 2009-12-12 at 16:44 -0500, Felix Miata wrote:
kernel lines with strings like root=/dev/disk/by-label/ata-ST3320620AS_6QF3PJXC-part10 in menu.lst shorter, maybe shorter enough to fit on one line instead of wrapping.
Just use an editor that does not wrap. What are you using? - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkskEP0ACgkQtTMYHG2NR9VmEACeNumoaaweUs4X0MBK6PT5L8Xx TCQAn3hkGNsGjbFmvHo+TcmLfsfUdXY7 =NWAJ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/12 22:53 (GMT+0100) Carlos E. R. composed:
On Saturday, 2009-12-12 at 16:44 -0500, Felix Miata wrote:
kernel lines with strings like root=/dev/disk/by-label/ata-ST3320620AS_6QF3PJXC-part10 in menu.lst shorter, maybe shorter enough to fit on one line instead of wrapping.
Just use an editor that does not wrap. What are you using?
FWIW, which is irrelevant here, most of my editing is in mcedit. Whether an editor can or does wrap misses the point. I cannot mentally deal with fstab or menu.lst lines that cannot be seen in their entirety all at once on a single line. /dev/disk/by-id & /dev/disk/by-uuid lines create the problem that I cannot do that. It less frequently occurs when using the shorter /dev/disk/by-label method, and rarely occurs using device names or the currently broken old LABEL= method. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Bob S wrote:
Thanks for replying. Seems you always come to my rescue.
Not just you... Thanks Felix, that was a great summary of bootloader changes that we should all be aware of. Learning has occurred :p -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
Any further instruction/guidance is certainly appreciated.
A reread of https://bugzilla.novell.com/show_bug.cgi?id=483136#c13 gave me another idea to try: remove the whole root= string from the Grub kernel line and see what happens. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/09 02:04 (GMT-0500) Bob S composed:
On Tuesday 08 December 2009 01:04:45 Felix Miata wrote:
On 2009/12/07 23:45 (GMT-0500) Bob S composed:
I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system)
Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic.
Boot 11.0 and find out the name of the partition on which 11.2 is installed, e.g. /dev/sda6. Then boot to the 11.2 Grub menu, and substitute that at the root= position. Once you have 11.2 booted you can repair the Grub menu and fstab if necessary.
Thanks for replying. Seems you always come to my rescue. The boot partition is indelibly pasted in my mind. (sdc8) Not sure what you mean about the "root=position". You mean as an option on the grub menu? You mean /dev/sdc8 ? or (hd2,7) ?
FYI every partition on every disk is mounted no matter what OS is mounted and run. I can see everything and access everything. I can look at the 11.2 /boot/grub/menu.lst or the fstab or whatever. I have made sure that every menu.lst is exactly the same for every OS.
That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software?
You're hitting fallout from widespread changes made to support booting from USB and other media that may or may not be present on any given boot. Apparently part of the disk label for that partition in /dev is tied to the controller on the dead motherboard, but its the label hiding in the initrd, not the applicable one, if I'm right about what's going on. It may mean you won't be able to boot using the old initrd and will have to rescue boot to build a new one, depending on whether the initrd supports the necessary device name.
OK, I have noticed that the new boot sequence is different from what I am used to seeing. Lots of USB stuff which is kind of confusing to me. So, can I build a new initrd for 11.2 ? You spoke of rescue boot for grub. Is that part of the grub command? Guess I will have to review my grub options.
When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please?
Exactly when was the old motherboard made? Make? Model? If it's several years old it could be victim of premature cap failure, nothing to do with actually using or misusing it.
The old motherboard was an MSI K8T with a 939 pin AMD processor, about 4 years old. Can't believe that the 939 is "passe" Was definitely caps. The new board is also MSI "V" series with a dual core AMD processor.
Any further instruction/guidance is certainly appreciated.
For those who were following this thread and/or trying to help, everybody seems to have been lead astray, by the fact that 11.0 would still boot, from asking a critical question: Do the chipsets in both the old and new motherboards require the same storage drivers? The answer turned out to be no. The reason 11.0 would boot while 10.2 and 11.2 would not is (AFAICT) that its /etc/sysconfig/kernel's INITRD_MODULES line specified both VIA (old chipset) and AMD (new chipset) storage drivers, while for 10.2 & 11.2 only those for VIA were specified, which meant that the existing initrds built for 10.2 and 11.2 were missing the drivers required to find any disks at all after the motherboard replacement. Bob brought his puter here to visit me about 11 hours ago, which is how we got this figured out. We stumbled around quite a while with Knoppix, 11.0 and DFSee boots thinking some kind of Grub and/or BIOS drive order trouble until eventually I realized looking at a Grub prompt's 'find /boot/grub/stage1' output that 11.2's root partition didn't seem to exist, which is when the light bulb clicked on. We built a new initrd for 11.2 by booting installed system from the 11.2 DVD's rescue menu. Then we built one for 10.2 via chroot from an 11.0 boot. When he left, the machine was able to boot 10.2, 11.0 & 11.2, all from using 11.0's Grub menu without any chainloading. The question remains why only 11.0 would have specified inclusion of drivers for one chipset that was not on the old motherboard (i.e. AMD), but not others (e.g. Intel). -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system)
Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic.
That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software?
When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please?
Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Hello: I would try to boot the openSUSE 11.0 using an openSUSE 11.0 install DVD. Select repair system at the beginning, and later select expert tools, then boot installed system. (I am not exactly sure these arec the correct terms you will see, but similar. ) Then all the available installed system should be recognized, and you have to select the one you want to boot. Choose 11.0 and boot it. You should be able to login 11.0 and you can use YAST to reinstall a boot loader. For multiple systems I prefer chainloading. Ie. the operation systems are loaded from their root partition's boot sectors, and the MBR boot sector just points to these boot sectors. For chainloading I use lilo, as I had no time yet to learn grub. I hope this helps. Cheers, Istvan -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tuesday 08 December 2009 05:38:16 Istvan Gabor wrote:
I installed 11.2 a couple of weeks ago and have been playing with it on and off. Booted and ran just fine. At the present I have 3 other OS's installed. (10.2, 11.0, and winders) (11.0 is my everyday user system)
Last thursday I had a catastrophic failure of my MB. I replaced the MB and processor and got up running again. Except that 11.2 won't boot now. It starts with grub and gets about half way through the boot process and fails with the message "Cannot locate /dev/disk/by-label/11.2". Then it offers to look for the /dev/disk/by-id (which it identifies correctly) and I answer yes. Then it grinds away for awhile generating several lines of dots, and then tries to dump me into a shell and also fails with a kernel panic.
That is the very disk it is actually booting from???? and it can't find it?? To be fair, the grub that it is using is on 11.0 There are no logs generated. And maybe I don't understand the boot process well enought. Does the process run for a long time before it gets around to loading the software?
When the failure occurred it was at turning the machine on. What could have happened to cause this and how do I fix it? Some clues, ideas please?
Bob S
Hello:
I would try to boot the openSUSE 11.0 using an openSUSE 11.0 install DVD. Select repair system at the beginning, and later select expert tools, then boot installed system. (I am not exactly sure these arec the correct terms you will see, but similar. ) Then all the available installed system should be recognized, and you have to select the one you want to boot. Choose 11.0 and boot it. You should be able to login 11.0 and you can use YAST to reinstall a boot loader.
Hello Istvan. Thanks for replying. I have done exactly as you have suggested. Both with the 11.0 disk and the 11.2 disk. And I have "booted the installed system", both 11.0 and 11.2 successfully, although it misidentifies the disk. When I modify the grub to reflect where everything is properly, I end up with an error message that it could not make the changes. Grrrr.....I hate that "rescue system". It never works properly for as long as I can remember.
For multiple systems I prefer chainloading. Ie. the operation systems are loaded from their root partition's boot sectors, and the MBR boot sector just points to these boot sectors. For chainloading I use lilo, as I had no time yet to learn grub.
Tried to chainload also. Didn't help. I used LILO years ago but when grub became the popular boot manager I changed to it.
I hope this helps.
Thanks for trying. It is appreciated. Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Thanks for replying. I have done exactly as you have suggested. Both with the 11.0 disk and the 11.2 disk. And I have "booted the installed system", both 11.0 and 11.2 successfully, although it misidentifies the disk. When I modify the grub to reflect where everything is properly, I end up with an error message that it could not make the changes.
OK, according to this you have a working system when booting with install DVD. I would boot eg. 11.2 using the DVD. Then I would move /boot/grub to /boot/grub.backup and would try to install/create boot loader from scratch in yast (select suggest new configuration). I would also check /etc/fstab and use consistent partition names according to /dev/disk/by-id entries both in /etc/fstab and grub files (just for safety). If this still gives error, then you probably have to install grub using command line. Or CTR+ALT+F10 may show the error messages given by YAST/Grub installer. Cheers, Istvan -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Wednesday 09 December 2009 15:02:27 Istvan Gabor wrote:
Thanks for replying. I have done exactly as you have suggested. Both with the 11.0 disk and the 11.2 disk. And I have "booted the installed system", both 11.0 and 11.2 successfully, although it misidentifies the disk. When I modify the grub to reflect where everything is properly, I end up with an error message that it could not make the changes.
OK, according to this you have a working system when booting with install DVD. I would boot eg. 11.2 using the DVD. Then I would move /boot/grub to /boot/grub.backup and would try to install/create boot loader from scratch in yast (select suggest new configuration). I would also check /etc/fstab and use consistent partition names according to /dev/disk/by-id entries both in /etc/fstab and grub files (just for safety).
Hello Istvan, I could do this of course, but I am afraid to. The stupid rescue system mis-identifies my hard drives. It calls sdc sda, so if I write it that way and then gets installed into the MBR I will really have big problems. I suppose I could rename the disks for the purposes of the Grub menu but that still worries me. It is strange though, in the menu of "Boot installed system" it calls 11.2 as being sda8 (when really it is sdc8) it still boots to the OS.
If this still gives error, then you probably have to install grub using command line. Or CTR+ALT+F10 may show the error messages given by YAST/Grub installer.
Dunno. Where do I go from here. CTR+ALT+F10 gives me nothing, just like the boot log. Bob S -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Hi Bob. I have a similar hardware set as you. Two harddisks with different OSs. I think we're suffering the same problem. The problem of mine is that when OpenSUSE installs, it's recognizes the disk as hd0. And after the installation, it's hd1. I've never solved this problem, no matter how much configs I changed, It always refused to work. Grub failed to load, or Grub loads, but failed to find the root device. My solution is to press F12(may be diffrent at your system, just to get to select the boot device, some machine use ESC or F2) after power on the machine, manually select the disk where OpenSUSE was installed. Hope this would helps. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 2009/12/11 15:00 (GMT+0800) Yang Bo composed:
Hi Bob. I have a similar hardware set as you. Two harddisks with different OSs. I think we're suffering the same problem.
Which operating systems are installed, and on which disks?
The problem of mine is that when OpenSUSE installs, it's recognizes the disk as hd0. And after the installation, it's hd1. I've never solved this problem, no matter how much configs I changed, It always refused to work. Grub failed to load, or Grub loads, but failed to find the root device.
Is one PATA and the other SATA? If yes, describe the settings in the BIOS for their controllers. Also, share with us the output of fdisk -l (or equivalent if not only Linux is installed) while booted to each OS. -- " We have no government armed with power capable of contending with human passions unbridled by morality and religion." John Adams, 2nd US President Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
OK, according to this you have a working system when booting with install DVD. I would boot eg. 11.2 using the DVD. Then I would move /boot/grub to /boot/grub.backup and would try to install/create boot loader from scratch in yast (select suggest new configuration). I would also check /etc/fstab and use consistent partition names according to /dev/disk/by-id entries both in /etc/fstab and grub files (just for safety).
Hello Istvan,
I could do this of course, but I am afraid to. The stupid rescue system mis-identifies my hard drives. It calls sdc sda, so if I write it that way and then gets installed into the MBR I will really have big problems. I suppose I could rename the disks for the purposes of the Grub menu but that still worries me.
OK, let's see. When you insert the install DVD, you have a menu. It has "Repair installed system" and "Rescue system" options. Which one did you choose? I suggest you going with "Repair installed system". Then you choose load installed system. This should boot your system (the one you choose) as normal. In this case CTRL+ALT+F10 should show differerent log messages. Lack of this indicates for me that you have not booted your system but you started the rescue mode (though I am not sure).
It is strange though, in the menu of "Boot installed system" it calls 11.2 as being sda8 (when really it is sdc8) it still boots to the OS.
This is the reason why I suggested naming the drives according to their ID both in /etc/fstab and grub boot loader. Probably this is done already if ou system is booted.
Dunno. Where do I go from here. CTR+ALT+F10 gives me nothing, just like the boot log.
See my above note. For now what I can say: you have to be sure that you boot the system on the hard drive when you use the install DVD. The you start yast and install new bootloader. Cheers, Istvan -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (7)
-
Bob S -
Carlos E. R. -
David C. Rankin -
Felix Miata -
Istvan Gabor -
Philipp Thomas -
Yang Bo