Re: [newbie] bad superblock

2004-03-06 Thread Josenildo Marques
On Sat, 2004-03-06 at 19:50, Brian Parish wrote:
 On Sun, 2004-03-07 at 08:54, Josenildo Marques wrote:
  Hello !
  
  I installed 9.2 for a friend of mine and everything went smoothly. I
  even updated it completely. His computer had a Windows partition and I
  decided to turn it into a /home partition. I think my mistake was that I
  didn't format it - I was running HardDrake. Now his computer won't start
  because of a superblock error - partition /dev/hda6 is said to have
  superblock errs.
  Is there a way I can (re)format it and create the /home directory again
  ? I was advised to boot in failsafe mode and run e2fsck, but
  unfortunately it didn't work.
  Please, I really need help.
  TIA
 
 You can boot into single use mode using failsafe as you did before, but
 you need to then format the partition.  Running e2fsck won't do that.
 
 Try: mkfs.ext3 /dev/hda6
 
 to create the default journalized filesystem on this partition.
 
 HTH
 Brian


Thank you very much, Brian. Everything is OK now.
 
 __

-- 
josenildo marques 
icq #289971493 
homepage http://cyb.ezdir.net
registered linux user #341648
*
Do not wait for the last judgment. It takes place every day. Albert
Camus


Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-28 Thread Hoyt Bailey

- Original Message - 
From: David E. Fox [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, February 27, 2004 21:00
Subject: Re: [newbie] Bad superblock


 On Fri, 27 Feb 2004 11:06:14 -0600
 Hoyt Bailey [EMAIL PROTECTED] wrote:


  lilo. This worked but when I booted into 10.0 there was an error which
  I couldnt clear or get around so master reset was pushed to get out of
  the system.  On rebooting to 9.2 several times I noticed scroll past
  on the

 Hoyt - this might be similar to my experience with 10.0. I'm on rc1
 (updated through urpmi) but my 9.2 partition on /dev/hda7 had (and may
 still have) serious problems booting successfully. There were
 indications that /etc/fstab on that partition was scrambled (zero bytes
 interspersed with text) and the boot loader (lilo) on /dev/hda isn't all
 that good compared to the boot loader on /dev/hdb that boots up 10.0.
 Moral - don't have two bootable partitions :) as it can get confusing.

 If rescue could simply boot hda7 from the CDROM it would be a lot easier
 to get into 9.2 and/or get the fix. Thomas Backlund posted some help
 recently, but I don't remember the exact subject, you might want to
 search the archives.

 Other distributions I have used in the past allow this by just typing
 'linux boot=/dev/hda7' on the command line of the rescue disk, but
 Mandrake doesn't allow this choice.

 I also noticed that the modules for the kernel were not accessible,
 which doesn't let me do all that much. I tried copying them over, but
 that didn't work either.

 And, I think you may be confusing the boot partition table with the
 superblock. Those are two completely different things.

  screen a note that said bad superblock and the boot was switched to
  10.0 by the computer now that is all I can boot into even though the
  boot process

 Well, in case it really is a bad superblock (have you mounted the
 partitions from rescue?) then e2fsck will fix it, and if there really is
 a damaged superblock, the filesystem (I assume for the moment you are
 using ext2) stores copies every 8192 blocks, with the first block
 starting at 1, so then the 2nd copy is at 8193, the third at 8193+8192,
 and so forth. If you type e2fsck -b 8193 /dev/hda (use the right
 partition) it should be able to clear things up.

 Once cooker stabilizes somewhat, I'm planning on moving it from
 /dev/hdb6 back to the original place, /dev/hdb7. simply because
 /dev/hdb7 is much bigger, and that gives me extra space to store things
 :). It would be kinda neat to just save the state of the system at
 present (i.e, all the rpms and their versions and so forth, and
 automatically grab any rpms that are in contrib. mklivecd may be what
 I'm looking for, but I haven't tried it yet.

David
Richard Urwin proposed a theory. With which I agree. I installed 10.0rc1 on
the second disk while it was the only disk in the system (hda).  Then
reinstalled the main disk which had 9.2 on (hda).  I then rebooted 10.0 and
the system had two OS's on apparantly the same disk  since the mount tables
both directed the system to use different inodes on hda it got confused and
did the best it could.  At the same time it drove me crazy (the hidden
purpose of computers).  Anyway I erased the slave disk and converted it to
fat32 to act as my music file.  This released 9.2 to act by its self and it
is indeed screwed up e2fsck repaired 6 filesystems, there shold only be 3.
I will reinstall 10.0 on the main disk and that should get me back to a sort
of reallity.  I wish I had a program to reduce the windows part to half its
size but I'll have to live with it.

Thanks;
Hoyt



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad Superblock

2004-02-28 Thread Richard Urwin
On Saturday 28 Feb 2004 6:35 am, Hoyt Bailey wrote:
 From: Richard Urwin [EMAIL PROTECTED]
  Howabout this for a theory:
  You installed 10.0 on the second disk while it was the only disk in
  the machine, and therefore was hda. Your mount table, which
  specifies which disk partitions are mapped to which Linux
  directories, refers to the partitions on disk b as if they are on
  disk a. You are therefore using the partitions on the 9.2 disk. The
  boot procedure works because you are booting from the second
  disk(?) or you have corrected the grub conf on hda to point to the
  10.0 kernel(?) So the kernel loads OK, but the mounted partitions
  do not point at the 10.0 files, they point at the 9.2 files. In
  trying to fix it, or in just running 10.0, you corrupted something
  in the 9.2 files so 9.2 doesn't work either.

 Ok thats a reasonable explanition. So the solution is to reformat hdb
 and reinstall 10.0 to hda. A question if I may is there an easy way
 to transfer windows to hdb (dosent sound like there is).  I cant
 reinstall windows because each time I have I have to spend 3 or 4
 days because viruses sneek in while downloading the virus
 definations. Thanks for your input Richard I think you are right.

Send us details of exactly how the two disks are partitioned - use 
diskdrake - and the contents of the two /etc/fstab files. I thought you 
said that you removed XP.

It may still be possible to rescue at least one disk.
I don't fancy your chances in moving Windows, but you shouldn't have to. 
If you don't re-partition the disks it can stay where it is.

-- 
Richard Urwin

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-28 Thread Richard Urwin
On Saturday 28 Feb 2004 4:36 pm, Hoyt Bailey wrote:
 I wish I had a program to reduce the windows part
 to half its size but I'll have to live with it.

The standard method is:
Defragment Windows, do not select optimise layout (or similar)
diskdrake now allows you to shrink the windows partition.

We will know more when we get details about which partitions you have 
and what's on them.

-- 
Richard Urwin

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad Superblock

2004-02-28 Thread Hoyt Bailey

- Original Message - 
From: Richard Urwin [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Saturday, February 28, 2004 04:39
Subject: Re: [newbie] Bad Superblock


 On Saturday 28 Feb 2004 6:35 am, Hoyt Bailey wrote:
  From: Richard Urwin [EMAIL PROTECTED]
   Howabout this for a theory:
   You installed 10.0 on the second disk while it was the only disk in
   the machine, and therefore was hda. Your mount table, which
   specifies which disk partitions are mapped to which Linux
   directories, refers to the partitions on disk b as if they are on
   disk a. You are therefore using the partitions on the 9.2 disk. The
   boot procedure works because you are booting from the second
   disk(?) or you have corrected the grub conf on hda to point to the
   10.0 kernel(?) So the kernel loads OK, but the mounted partitions
   do not point at the 10.0 files, they point at the 9.2 files. In
   trying to fix it, or in just running 10.0, you corrupted something
   in the 9.2 files so 9.2 doesn't work either.
 
  Ok thats a reasonable explanition. So the solution is to reformat hdb
  and reinstall 10.0 to hda. A question if I may is there an easy way
  to transfer windows to hdb (dosent sound like there is).  I cant
  reinstall windows because each time I have I have to spend 3 or 4
  days because viruses sneek in while downloading the virus
  definations. Thanks for your input Richard I think you are right.

 Send us details of exactly how the two disks are partitioned - use
 diskdrake - and the contents of the two /etc/fstab files. I thought you
 said that you removed XP.

 It may still be possible to rescue at least one disk.
 I don't fancy your chances in moving Windows, but you shouldn't have to.
 If you don't re-partition the disks it can stay where it is.

 -- 
 Richard Urwin

No I didnt remove windows yet I have to use it to talk to you for now. I did
erase the slave and converted it to fat32.  I will use it for my music
partition. While I would like to reduce the windows part by half I dont have
anything that would do that and I cant take the chance of trashing it right
now.  I still have 65GB that should be enough to get 10.0 running.  The 9.2
installation is trashed I ran e2fsck -b 6 times and the way I figure it
there should be only 3 fs on the disk with windows 4 but it shoulnt touch a
NTFS.  I will reinstall 10.0rc1 on the main disk, as soon as I figure out
what part's I want to create. This on balance was a learning experience and
should teach me to do it right the first time.  Funny it hasent worked yet.
Thanks for your theory it appears to be correct.

Regards;
Hoyt



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-28 Thread Hoyt Bailey

- Original Message - 
From: Richard Urwin [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Saturday, February 28, 2004 04:42
Subject: Re: [newbie] Bad superblock


 On Saturday 28 Feb 2004 4:36 pm, Hoyt Bailey wrote:
  I wish I had a program to reduce the windows part
  to half its size but I'll have to live with it.
 
 The standard method is:
 Defragment Windows, do not select optimise layout (or similar)
 diskdrake now allows you to shrink the windows partition.
 
 We will know more when we get details about which partitions you have 
 and what's on them.
 
 -- 
 Richard Urwin
 
Sorry partition 
Disk hda
#1 Windows XP(NTFS 46.57GB)
#2  Boot (for 9.2 63MB)
#3  Root(9.2 64.65GB)
#4  Swap(518MB)
Disk hdb
#1 Music(formatted as fat32 27.96GB)

Regards;
Hoyt


Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-28 Thread Richard Urwin
On Saturday 28 Feb 2004 5:52 pm, Hoyt Bailey wrote:
 Disk hda
 #1 Windows XP(NTFS 46.57GB)
 #2  Boot (for 9.2 63MB)
 #3  Root(9.2 64.65GB)
 #4  Swap(518MB)
 Disk hdb
 #1 Music(formatted as fat32 27.96GB)

I would:
1) copy anything you don't want to lose (like the contents of /home) to 
Music.
2) install 10.0 using the existing partitions, but reformatting them.

when you're happy everything is stable and usable:
3) backup anything from Windows that you don't want to lose to Music.
4) defragment Windows. Do not optimise the layout.
5) boot linux from CD
6) shrink the windows partition
7) create a new partition at the end of Windows
8) that may rename all your partitions*, so edit /etc/fstab to suit.

*I would guess that you have a single primary partition and an extended 
partition on hda, making Boot hda5. If so creating a new primary 
partition will not rename anything.
-- 
Richard Urwin

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-28 Thread Hoyt Bailey

- Original Message - 
From: Richard Urwin [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Saturday, February 28, 2004 07:12
Subject: Re: [newbie] Bad superblock


 On Saturday 28 Feb 2004 5:52 pm, Hoyt Bailey wrote:
  Disk hda
  #1 Windows XP(NTFS 46.57GB)
  #2  Boot (for 9.2 63MB)
  #3  Root(9.2 64.65GB)
  #4  Swap(518MB)
  Disk hdb
  #1 Music(formatted as fat32 27.96GB)

 I would:
 1) copy anything you don't want to lose (like the contents of /home) to
 Music.
 2) install 10.0 using the existing partitions, but reformatting them.

 when you're happy everything is stable and usable:
 3) backup anything from Windows that you don't want to lose to Music.
 4) defragment Windows. Do not optimise the layout.
 5) boot linux from CD
 6) shrink the windows partition
 7) create a new partition at the end of Windows
 8) that may rename all your partitions*, so edit /etc/fstab to suit.

 *I would guess that you have a single primary partition and an extended
 partition on hda, making Boot hda5. If so creating a new primary
 partition will not rename anything.
 -- 
 Richard Urwin

I can do everything except 6, just dont know how.  What is the new partition
at the end of windows for [win swap, if so how about moving swap there and
let both systems use it windows identifies the current swap partition as a
logical part. so it should be able to use it].  I havent caught linux using
swap yet but when I get on spreadsheets that may change.

Regards;
Hoyt



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-27 Thread Hoyt Bailey

- Original Message - 
From: Raffaele Belardi [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, February 27, 2004 01:42
Subject: Re: [newbie] Bad superblock


 Assuming you are using ext2 or ext3, superblock backup copies are stored
 in several places on the disk, mke2fs tells you where when you first
 create the filesystem.

 If you didn't take a note, you can find out where the backup copies are
 by typing
 # mke2fs -n /dev/hdxx
 (note -n, does not actually create the fs, only prints what it would do).

 Once you know where the backups are, you can issue
 # e2fsck -b block number /dev/hdxx
 to force fsck to use a different, possibly uncorrupted, superblock.

 All of this can normally be done with the rescue option of the MDK9.2
 install disk (i.e. insert disk 1, reboot, hit F1, type rescue), I'm
 surprised that it does not work for you. Could you please give more
 details on your problem?


Well I'll try. I installed 10.0rc1 and missed the configuration of the mouse
although the mouse worked fine during the install, it didnt register that I
hadnt configujred it, when I booted into the new system the mouse wouldnt
even move or otherwise respond otherwise the install went well.  I decided
to reinstall and this time I configured the mouse.  On booting into the
system the mouse worked but there was another problem that requried me to
correct lilo in order to boot 10.0.  Well I misspelled lilo and couldnt find
lilo.conf.  Therefore I figured the easy way to fix lilo would be to
reinstall with everything in place so I did so and selected grub instead of
lilo. This worked but when I booted into 10.0 there was an error which I
couldnt clear or get around so master reset was pushed to get out of the
system.  On rebooting to 9.2 several times I noticed scroll past on the
screen a note that said bad superblock and the boot was switched to 10.0 by
the computer now that is all I can boot into even though the boot process
starts to boot into 9.2 it always ends up in 10.0 with the unresolvable
problem.  Rescue boots ok but fsck command will not execute and that is the
first time I have tried to use rescue so I am at a rather sever impass.
Hope this helps.

Regards;
Hoyt



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-27 Thread Raffaele Belardi
Hoyt,

Let me recap to see if I understand correctly:

- the grub menu presents you with two installations, 9.2 and 10.0
- when you select 10.0 the boot begins but for some reason it hangs at 
some point
- when you select 9.2 grub loads 10.0 anyway (and ends up as above)

I do not think the superblock error you see is related to the above, 
looks more like when you installed grub something was not done right and 
you ended up with two entries linking to the same kernel image (10.0). 
The loader I use (lilo) does not have the intelligence necessary to do 
the switch you suggest, I assume it is the same for grub.

You could try to use the reinstall boot loader of the rescue disk:
1. boot from the MDK9.2 cd1
2. at the splash, type F1, then rescue
3. at the menu, select reinstall boot loader
This should search for existing boot images on the disk and reconstruct 
the lilo (or grub) loader. But, I have never done it so I cannot tell 
you how effective it is.

Another thing you can try is to select mount existing partitions under 
/mnt at point 3. above. This will let you access the uncorrupted 
partitions from the rescue disk, and then you could try to modify fstab 
to avoid loading the corrupt partition. But this option depends on how 
familiar you are with the command line, and with the linux configuration 
files.

I don't remember where fsck is located in the rescue disk, but I am 
almost sure I had e2fsck in the path. Try it out, it's the same as fsck 
if you are using ext2 or ext3 (the default installation).

It's not much, I hope others have better ideas.

raffaele

[EMAIL PROTECTED] wrote:
- Original Message - 
From: Raffaele Belardi [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, February 27, 2004 01:42
Subject: Re: [newbie] Bad superblock



Assuming you are using ext2 or ext3, superblock backup copies are stored
in several places on the disk, mke2fs tells you where when you first
create the filesystem.
If you didn't take a note, you can find out where the backup copies are
by typing
# mke2fs -n /dev/hdxx
(note -n, does not actually create the fs, only prints what it would do).
Once you know where the backups are, you can issue
# e2fsck -b block number /dev/hdxx
to force fsck to use a different, possibly uncorrupted, superblock.
All of this can normally be done with the rescue option of the MDK9.2
install disk (i.e. insert disk 1, reboot, hit F1, type rescue), I'm
surprised that it does not work for you. Could you please give more
details on your problem?

Well I'll try. I installed 10.0rc1 and missed the configuration of the mouse
although the mouse worked fine during the install, it didnt register that I
hadnt configujred it, when I booted into the new system the mouse wouldnt
even move or otherwise respond otherwise the install went well.  I decided
to reinstall and this time I configured the mouse.  On booting into the
system the mouse worked but there was another problem that requried me to
correct lilo in order to boot 10.0.  Well I misspelled lilo and couldnt find
lilo.conf.  Therefore I figured the easy way to fix lilo would be to
reinstall with everything in place so I did so and selected grub instead of
lilo. This worked but when I booted into 10.0 there was an error which I
couldnt clear or get around so master reset was pushed to get out of the
system.  On rebooting to 9.2 several times I noticed scroll past on the
screen a note that said bad superblock and the boot was switched to 10.0 by
the computer now that is all I can boot into even though the boot process
starts to boot into 9.2 it always ends up in 10.0 with the unresolvable
problem.  Rescue boots ok but fsck command will not execute and that is the
first time I have tried to use rescue so I am at a rather sever impass.
Hope this helps.
Regards;
Hoyt






Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-27 Thread Hoyt Bailey

- Original Message - 
From: Raffaele Belardi [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Friday, February 27, 2004 08:44
Subject: Re: [newbie] Bad superblock


 Hoyt,

 Let me recap to see if I understand correctly:

 - the grub menu presents you with two installations, 9.2 and 10.0
 - when you select 10.0 the boot begins but for some reason it hangs at
 some point

No.  It dosent hang but completes the boot into KDE (via login).  KDE has a
problem with something that will not resolve or clear.  It just comes back
when it is closed(I forgot what it is). Cancle also dosent and there is not
enough time to click on anything else before it reappears.

 - when you select 9.2 grub loads 10.0 anyway (and ends up as above)

Yes,  although it starts to load 9.2,  you can tell because the background
has a 9.2 on it but it ends up with the 10.0 KDE.

 I do not think the superblock error you see is related to the above,
 looks more like when you installed grub something was not done right and
 you ended up with two entries linking to the same kernel image (10.0).
 The loader I use (lilo) does not have the intelligence necessary to do
 the switch you suggest, I assume it is the same for grub.

 You could try to use the reinstall boot loader of the rescue disk:
 1. boot from the MDK9.2 cd1
 2. at the splash, type F1, then rescue
 3. at the menu, select reinstall boot loader
 This should search for existing boot images on the disk and reconstruct
 the lilo (or grub) loader. But, I have never done it so I cannot tell
 you how effective it is.

 Another thing you can try is to select mount existing partitions under
 /mnt at point 3. above. This will let you access the uncorrupted
 partitions from the rescue disk, and then you could try to modify fstab
 to avoid loading the corrupt partition. But this option depends on how
 familiar you are with the command line, and with the linux configuration
 files.

 I don't remember where fsck is located in the rescue disk, but I am
 almost sure I had e2fsck in the path. Try it out, it's the same as fsck
 if you are using ext2 or ext3 (the default installation).


This may complicate things a bit.  9.2 is a Reiserfs installation.  I think
10.0 was installed as a ext2 or ext3.  I'm not absolutely sure I didnt have
enough time to look.

 It's not much, I hope others have better ideas.

 raffaele

The only reason I was trying to solve the superblock problem was because
that was the only handle I had.  Would a reiserfs partion make any
difference to your suggestions?

Regards;
Hoyt



Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


Re: [newbie] Bad superblock

2004-02-27 Thread David E. Fox
On Fri, 27 Feb 2004 11:06:14 -0600
Hoyt Bailey [EMAIL PROTECTED] wrote:


 lilo. This worked but when I booted into 10.0 there was an error which
 I couldnt clear or get around so master reset was pushed to get out of
 the system.  On rebooting to 9.2 several times I noticed scroll past
 on the

Hoyt - this might be similar to my experience with 10.0. I'm on rc1
(updated through urpmi) but my 9.2 partition on /dev/hda7 had (and may
still have) serious problems booting successfully. There were
indications that /etc/fstab on that partition was scrambled (zero bytes
interspersed with text) and the boot loader (lilo) on /dev/hda isn't all
that good compared to the boot loader on /dev/hdb that boots up 10.0.
Moral - don't have two bootable partitions :) as it can get confusing. 

If rescue could simply boot hda7 from the CDROM it would be a lot easier
to get into 9.2 and/or get the fix. Thomas Backlund posted some help
recently, but I don't remember the exact subject, you might want to
search the archives.

Other distributions I have used in the past allow this by just typing
'linux boot=/dev/hda7' on the command line of the rescue disk, but
Mandrake doesn't allow this choice. 

I also noticed that the modules for the kernel were not accessible,
which doesn't let me do all that much. I tried copying them over, but
that didn't work either.

And, I think you may be confusing the boot partition table with the
superblock. Those are two completely different things. 

 screen a note that said bad superblock and the boot was switched to
 10.0 by the computer now that is all I can boot into even though the
 boot process

Well, in case it really is a bad superblock (have you mounted the
partitions from rescue?) then e2fsck will fix it, and if there really is
a damaged superblock, the filesystem (I assume for the moment you are
using ext2) stores copies every 8192 blocks, with the first block
starting at 1, so then the 2nd copy is at 8193, the third at 8193+8192,
and so forth. If you type e2fsck -b 8193 /dev/hda (use the right
partition) it should be able to clear things up.

Once cooker stabilizes somewhat, I'm planning on moving it from
/dev/hdb6 back to the original place, /dev/hdb7. simply because
/dev/hdb7 is much bigger, and that gives me extra space to store things
:). It would be kinda neat to just save the state of the system at
present (i.e, all the rpms and their versions and so forth, and
automatically grab any rpms that are in contrib. mklivecd may be what
I'm looking for, but I haven't tried it yet.


 Hoyt

-- 

David E. Fox  Thanks for letting me
[EMAIL PROTECTED]change magnetic patterns
[EMAIL PROTECTED]   on your hard disk.
---

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com


[newbie] Bad superblock

2004-02-26 Thread Hoyt Bailey
Could someone please tell me in detail how to repair a superblock with a
rescue disk in Mandrake 9.2.  I tried to run fsck with the rescue disk but
that didnt work (command not found).

Regards;
Hoyt




Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com