PDA

View Full Version : [ubuntu] 12.04 won't boot to desktop after install -- uuid error



Mark Phelps
May 2nd, 2012, 08:38 PM
Have been installing Ubuntu versions on this same PC since 11.04 and never encountered this failure ...

Installed from USB. Went fine. Removed USB rebooted. Never get to login screen.

Get error message of /dev/disk/by-uuid xxx does not exist

Ends with (intramfs) prompt

So, did blkid command -- uuid of Ext4 partition EXACTLY matches the uuid in the error message!

Erased partition, reinstalled again -- this time, from a CD. Rebooted -- same result.

Tried adding nomodeset to kernel boot parms -- no difference.

So basically, I can boot into LiveCD and LiveUSB just fine, but after install, can't get by the (intramfs) prompt.

Booted into 11.10 (where I am now) and ran CHECK on Ext4 filesystem on other disk -- the one containing Precise. Found no errors. Rebooted -- same problem.

Booted from Boot-Repair CD, downloaded update, selected option to repair -- just sits there with the activity thing moving. Left for over an hour, no progress.

OK, so now what?

darkod
May 2nd, 2012, 08:46 PM
I can't explain it, but it seems grub2 is not always installed properly when there is already a grub2 on the MBR from previous version.

It sounds silly, but try reinstalling grub2 with the standard commands from live mode (using the same version cd as the one you have installed on the hdd):
sudo mount /dev/sdXY /mnt
sudo grub-install --root-directory=/mnt /dev/sdX

If that doesn't help, try with parameter 'rootdelay=30' although that's mostly for usb disks.

Mark Phelps
May 2nd, 2012, 09:06 PM
Sorry ... been messing around with alternatives, trying to get it working ...

I booted into 11.10, ran update-grub, and tried booting Precise from there -- same problem

But, when I chose the Older kernel versions, specifically, kernel 3.2.0-23, it boots up just fine!

So WHAT I discovered is that the latest kernel update is the problem!

When I choose the default (3.2.0-24), it fails to boot.

And this is using the GRUB version installed with 11.10.

When I did the install, in both cases, I chose the option to include upgrades in the install -- this has been a GOOD thing in the past. Guess not, these days.

So, for now, I have a workaround.

When I get a chance, I'll change the BIOS to boot from the Precise drive and see if selecting the older kernel version works from there as well.

Mark Phelps
May 4th, 2012, 01:53 PM
I was WRONG!! Tried today booting into 12.04 using the older kernel and got (intramfs) again!

And, once again, Recovery Mode fails to get to a prompt.

I will try the rootdelay and, if that doesn't work, reinstalling GRUB -- but if those fail, this could end my activities with Ubuntu.

UPDATE: OK, tried rootdelay, reinstalling GRUB, even reinstalling 12.04 -- no improvement.

I've been installing and configuring Ubuntu since the 7.04 days, and I've encountered a few problems along the way -- but NONE with the severity of this version.

Maybe one of these days, Canonical will focus on getting Ubuntu to WORK -- and stop focusing on eye-candy (HUD) and Cloud (Ubuntu One).

johnmudd
May 7th, 2012, 06:53 PM
I've got the same problem. Any suggestions?

Morbius1
May 7th, 2012, 07:17 PM
@Mark Phelps,

Get error message of /dev/disk/by-uuid xxx does not existThere is an unbelievably old bug that used to produce that error. The only fix was to modify the fstab entry. Instead of specifying the device this way:

UUID=xxx-xxx-xxx /mountpoint ext4 defaults 0 0It had to be defined like this:

/dev/disk/by-uuid/xxx-xxx-xxx /mountpoint ext4 defaults 0 0Don't know if it's the same bug or if it's the same cause but ...

predato
May 7th, 2012, 08:35 PM
I had the similar problem yesterday, after a program installation I couldn't get login. Unfortunately I had to format my comp. I'm new to Ubuntu. It's annoying to stay helpless. I hope it gets solved soon.

Mark Phelps
May 7th, 2012, 09:38 PM
@Mark Phelps,
There is an unbelievably old bug that used to produce that error. The only fix was to modify the fstab entry. Instead of specifying the device this way:
It had to be defined like this:
Don't know if it's the same bug or if it's the same cause but ...

Thanks ... I've done a lot of other checking as well, including looking at the grub.cfg file to see if the Linux boot line is written in error. The actual error message does mention "dev/disk/by-uuid/xxx-xxx-xxx" and I checked to confirm that the UUID info is correct. So, there do not appear to be any errors in the grub.cfg file.

So, as another pass, I pulled out an old SATA drive, wiped it, formatted it (using GParted LiveCD) -- and installed to that. And that worked OK ... for a day.

And then, I started getting disk errors.

If the earlier IDE drive (containing 10.04) and the newer SATA drive (now containing 12.04) are now BOTH bad .. it looks like 12.04 could be a "drive-killer"...

So now, I'm wiping the SATA drive and will try again with another newer SATA drive. I'm expecting that the install will go OK. I'm hoping that it lasts more than one day, though, with 12.04.

vVvSHADOWvVv
May 7th, 2012, 09:41 PM
@Mark Phelps,
There is an unbelievably old bug that used to produce that error. The only fix was to modify the fstab entry. Instead of specifying the device this way:
It had to be defined like this:
Don't know if it's the same bug or if it's the same cause but ...

@Morbius1

This is incorrect, you must edit the grub device line from the initial grub menu upon boot and remove a specific attribute.

Do change anything in the fstab, you will hurt your system as the uuid is unique to your drive..

WARNING.

Do not do this.

Mark Phelps
May 7th, 2012, 09:41 PM
I had the similar problem yesterday, after a program installation I couldn't get login. Unfortunately I had to format my comp. I'm new to Ubuntu. It's annoying to stay helpless. I hope it gets solved soon.

You don't format a computer; you format a drive -- and I already did that and it did not help. What worked for me, although only for a day, was using an entirely different physical drive.

So basically, there is no fix for this -- at least, apart from replacing the drive.

darkod
May 7th, 2012, 09:47 PM
There was a similar issue long time ago and I haven't seen it lately so I forgot the exact syntax of the error.

I had the link bookmarked but it was a sourceforge page and they seem to be doing some maintenance right now. I'll post the link as soon as they are up (and I remember to check :) ).

It was something like commenting out few lines around the "search for uuid" part of /etc/default/grub if I remember correctly, then running update-grub to create new grub.cfg that will not search for that uuid.

That fix was for cases when the uuid is correct, like in your case.

johnmudd
May 7th, 2012, 09:59 PM
First a correction. I'm running Ubuntu 11.10, not 12.04. Sorry, I got my machines mixed up.

Apparently I had two problems. Here are my "solutions":


Ended up editing /boot/grub/grub.cfg, added rootdelay=90
That seems to eliminate uuid error. (I know, I shouldn't edit that file. Now I have to go back and edit /etc/grub.d/10_linux, add the rootdelay there and run "sudo update-grub" to properly generate a new grub.cfg file.)

linux /boot/vmlinuz-3.0.0-19-generic-pae root=UUID=ed8e5bb1-0cca-486c-a6b3-0985e1ceb45c ro rootdelay=90 quiet splash vt.handoff=7

Then it could find the drive but I was still stuck in the purple screen of death.

I used Alt-F1 to login, tried "sudo startx" but complained nvidia driver not available. I remembered Nvidia was listed in the recent Ubuntu updates I installed so it made me wonder if that was another problem.

So rebooted while holding Shift key,
Selected the next previous kernel from Grub menu,
Was able to boot into previous kernel in GUI mode.
Ran the "Additional Drivers" app from Dash.
Reselected the *Recommeded* Nvidia driver.
Rebooted into the standard kernel and I seem to be back to normal.

vVvSHADOWvVv
May 7th, 2012, 10:02 PM
First a correction. I'm running Ubuntu 11.10, not 12.04. Sorry, I got my machines mixed up.

Apparently I had two problems. Here are my "solutions":


Ended up editing /boot/grub/grub.cfg, added rootdelay=90
That seems to eliminate uuid error. (I know, I shouldn't edit that file. Now I have to go back and edit /etc/grub.d/10_linux, add the rootdelay there and run grub-mkconfig to properly generate a new grub.cfg file.)

linux /boot/vmlinuz-3.0.0-19-generic-pae root=UUID=ed8e5bb1-0cca-486c-a6b3-0985e1ceb45c ro rootdelay=90 quiet splash vt.handoff=7

Then it could find the drive but I was still stuck in the purple screen of death.

I used Alt-F1 to login, tried "sudo startx" but complained nvidia driver not available. I remembered Nvidia was listed in the recent Ubuntu updates I installed so it made me wonder if that was another problem.

So rebooted while holding Shift key,
Selected the next previous kernel from Grub menu,
Was able to boot into previous kernel in GUI mode.
Ran the "Additional Drivers" app from Dash.
Reselected the *Recommeded* Nvidia driver.
Rebooted into the standard kernel and I seem to be back to normal.

This issue is because the driver is being loaded at the improper time. You must completely purge nvidia from the system and instal the driver WITHOUT X running. It's a long painstaking process as I have been through this. I suggest you upgrade to 12.04 as it will solve most nvidia problems.

ṤⱧǠᴆŐƜ
shadowsgovernment.com

Oldnews2
May 16th, 2012, 05:55 AM
edit #2
lol :lolflag:
... the original uuid error is caused by the extensive file system checks involved in both ext3 and ext4 journaled file systems. if you are using "old reliable hard disks" you will probably get this error. the work around is to change the file system to ext2 when you install.

at least that was the only solution I have found that works on old reliable hard disks!

--------------------------------



Wow ... I'd like to "me to" that problem!

The following is the solution I found while googling ...

Here is the copy and pasted post from the crunchbang linux forum in it's entirety ...
Oh, the forum user's name was xaos52 so all credit is his not mine.


How to chroot from live session to #! on disk

Assumption: #! is installed on /dev/sda1 but is not accessible via grub
Goal: reinstall grub in mbr of /dev/sda:
Here goes:
boot from livecd or usb
in terminal window:

TARGET=/media/sda1
sudo mkdir -p $TARGET
sudo mount /dev/sda1 $TARGET
sudo mount --bind /dev $TARGET/dev
sudo mount --bind /dev/pts $TARGET/dev/pts
sudo mount --bind /proc $TARGET/proc
sudo mount --bind /sys $TARGET/sys
sudo chroot $TARGET /bin/bash

This places you in a root terminal on the #! system
Do whatever you want to do there, e.g. re-install grub

grub-install /dev/sda
update-grub

To exit from chroot:

control-D

This brings you back to where you were in the live session and you can unmount

sudo umount -l $TARGET/dev/pts
sudo umount -l $TARGET/dev
sudo umount -l $TARGET/proc
sudo umount -l $TARGET/sys

Thats it...
Reboot into the #! system on disk.

Edit1: Replaced '/usr/bin/bash' by '/bin/bash'
Edit2: Start with defining TARGET, and reference it in all subsequent lines.

Last edited by xaos52 (2012-03-26 16:30:46)
If you poke the bear it is going to come after you.

If you get an error message that says grub is not found and apt-get or aptitude can't find it (as I did), open up pcmanfm. In the already open chroot terminal:
nano /etc/apt/sources.listIn pcmanfm locate and open the live file system version of the same file. Copy and paste the repositories to the chrooted terminal apt source.list.

Than in the terminal,

apt-get update
ignore the notes on duplicate repos ... you can fix that later.
apt-get install grub2
you are done! reboot and enjoy!

thanx xaos52!

there may be another much easier fix ... I still get the same error message but not on all startups or reboots. I've had to basically use recovery mode to fix grub almost every other boot. As in one good boot, one bad boot with the original error message followed by recovery mode boot to fix grub followed by one good boot. grub bug anyone ???