[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

e2fsck says it fixed corrupted file system but it hasn't



Hello,


Thanks for your suggestions; so far, nothing has helped. Here
are your suggestions and what I've found:




>>  So I log in as root and run e2fsck using a spare superblcok:
>
>Have you tried running e2fsck with the default superblock manually?
>

It says

	warning: inode bitmap 196881 for group 23 not in group.
	Continue?<y>

	/dev/sdb2 is clean, no check




>
>>  So then I shutdown and reboot, thinking I've solved the 
>>  problem, but NO! When I reboot, I get exactly the same problem. 
>
>Why does it check the file system *at all* after a clean reboot?!

I don't know.


>
>>  So my question is, how come e2fsck says it fixed my problem,
>>  but it seems to not have?
>
>What if you run e2fsck again immediately after it's finished fixing the
>partition?

Good question; I get exactly the same result i.e. it finds the same
corruption and claims to fix it again.



>
>>  
>>  Here is my /etc/fstab, in case this is useful:
>>  
>>  /dev/sdb1   swap    swap     defaults   1   1
>>  /dev/sda1   /       ext2     defaults   1   1
>>  none        /proc   proc     defaults   1   1
>>  /dev/sdb2   /u1     ext2     defaults   1   1
>
>It should look like
>
>/dev/sdb1   swap    swap     defaults   0   0
>/dev/sda1   /       ext2     defaults   1   1
>none        /proc   proc     defaults   0   0
>/dev/sdb2   /u1     ext2     defaults   1   2
>

I made this change, but it seemed to have no effect whatsever.



>(BTW, when you swap the disks, you don't forget to adjust SCSI termination
>accordingly, do you?)

Another good question but no, this is not a problem, as I don't change
the physical location of the disks on the SCSI cable, but just swap
a jumper that sets SCSI ID.



>>
>> >  So I log in as root and run e2fsck using a spare superblcok:
>> >  So then I shutdown and reboot, thinking I've solved the 
>> >  problem, but NO! When I reboot, I get exactly the same problem. 
>
>Just turn the system to the 1 runlevel ( telinit 1 or as you wish ) ,
>repair the file system by fsck ,as you done, and do 'halt -n' - it's
>important ! 

I had great hopes that this would solve my problem, but it doesn't.
I tried many times, but doing the halt -n as you suggest seems to have
no effect. I did try this at run level 1.




So I'm still stuck.

Now I"ve heard that there is a program called "debugfs" which might
help me but could also do a lot of damage if it is not understood.
I searched around but can't find any man page, FAQ, readme, or any
other documentation on this dangerous program. Anyone out there have
any ideas?



Here for reference is my original email:

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Hiya folks,

I have a corrupted superblock.




System:

Old motherboard based on an Intel 486DX
16 Megs ram
Bustek SCSI controller, I think model 742
Two identical disk drives: DEC DSP3053LS




The disk drives are SCSI IDs 0 and 1.

I have enough disk space that I never really
used the second disk.

When I upgraded to Slackware 96 I simply 
swapped the SCSI IDs on the disks kept the old
disk as a backup of sorts.

Currently the lower drive is SCSI ID 0 and it
is the one I use.

This lower drive developed a corrupted superblock,
so I figured I'd swap the disks again, boot up my old
system from the upper drive, and fix the lower drive.

When I boot from the old upper drive, this is what it
finds on the lower drive (which is now sdb2):

===========================================================

Warning: inode bitmap 196881 for group 23 not in group

/dev/sdb2: unexpected inconsistency; run fsck manually

*************************************************************
fsck returned error code - reboot now!
*************************************************************

(after a minute or so it times out and goes to run level 
5 I think and says more:)

ext2-fs error (device 8/18): ext2-check-descriptors: inode
bitmap for group 23 not in group (block 196881)!

ext2-fs: group descriptors corrupted!

===========================================================


So I log in as root and run e2fsck using a spare superblcok:

e2fsck -b 8193 /dev/sdb2

and I get:

===========================================================

warning: inode bitmap 196881 for group 23 not in group.
Continue<y> (I continue)

Pass 1: checking inodes, blocks, and sizes
relocating group 23's inode bitmap to 196881
Pass 2:....
Pass 3:....
Pass 4:....
Pass 5:....
Fix summary information<y>? (I say yes)

Inode bitmap differences: (a whole bunch of numbers with 
plus signs in front of them) FIXED

/dev/sdb2: file system was modified

===========================================================

So then I shutdown and reboot, thinking I've solved the 
problem, but NO! When I reboot, I get exactly the same problem. 

So my question is, how come e2fsck says it fixed my problem,
but it seems to not have? Am I running e2fsck incorrectly?
Or is there some other problem?



Here is my /etc/fstab, in case this is useful:

/dev/sdb1   swap    swap     defaults   1   1
/dev/sda1   /       ext2     defaults   1   1
none        /proc   proc     defaults   1   1
/dev/sdb2   /u1     ext2     defaults   1   1









Thanks again for all your suggestions,

Michael Shiloh