Can resize_reiserfs do non-destructive resizing??

Can resize_reiserfs do non-destructive resizing??

[Bruce]

I have a desktop machine that was running Debian 3.0, installed on a 40Gig 
harddrive (single partition hda2 + swap hda1) formatted with reiserfs; the 
drive had about 10gig data on it. I wanted to test out Mandrake 9.0, and 
understood that the diskdrake program would handle resizing of the existing 
reiser partition (using resize_reiserfs, I assume). The diskdrake program 
suggested I backup as a precaution, but did not indicate that the resizing 
would be destructive.

I let it resize the existing partition to 20Gig. It seemed to work, and 
Mandrake installed without any problems. However, since then I have not been 
able to access the Debian partition - can't boot from it, can't mount it, 
nothing. Trying to boot from that partition results in:

 attempt to access beyond end of device
 03:02 rw=0 want 21364740 limit 21310191
 reiserfs_read_super: unable to read bitmap
 kernel panic....

I posted a query to alt.os.linux.mandrake, and got the answer that 
resize_reiserfs cannot do non-destructive repartitioning, and that my 
partition was toast (for the thread, search for the subject "Problem resizing 
reiserfs partition with DiskDrake" in alt.os.linux.mandrake from 10 November 
2002). I had a look at the man page, and if resize_reiserfs cannot, in fact, 
non-destructively resize a partition, I couldn't find anything saying so. 

Is this a problem with diskdrake's handling of resize_reiserfs?? 
resize_reiserfs itself??

It could also be due to physical problems with the hard drive, which had 
reported bad blocks some time ago. Is this more likely the cause of the 
problem??

Any thoughts,

Bruce

Re: Can resize_reiserfs do non-destructive resizing??

[Oleg Drokin]

Hello!

On Sun, Nov 17, 2002 at 09:40:49PM -0500, Bruce wrote:

> I have a desktop machine that was running Debian 3.0, installed on a 40Gig 
> harddrive (single partition hda2 + swap hda1) formatted with reiserfs; the 
> drive had about 10gig data on it. I wanted to test out Mandrake 9.0, and 
> understood that the diskdrake program would handle resizing of the existing 
> reiser partition (using resize_reiserfs, I assume). The diskdrake program 
> suggested I backup as a precaution, but did not indicate that the resizing 
> would be destructive.

resize_reiserfs is a non-destructive resizer.

> I let it resize the existing partition to 20Gig. It seemed to work, and 
> Mandrake installed without any problems. However, since then I have not been 
> able to access the Debian partition - can't boot from it, can't mount it, 
> nothing. Trying to boot from that partition results in:
>  attempt to access beyond end of device
>  03:02 rw=0 want 21364740 limit 21310191
>  reiserfs_read_super: unable to read bitmap
>  kernel panic....

Well, that means that the size of your debian partition is 10655092 megabytes,
but reiserfs filesystem is expecting the partition to be 10682370 megabytes
(as written in superblock).
You might try to use reiserfsck --rebuild-sb followed by subsequent
reiserfsck --rebuild-tree to bring reiserfs superblock data back in sync
with partition table. That will be dangerous if beginning of partition have
changed its location since mkfs time, though.

> Is this a problem with diskdrake's handling of resize_reiserfs?? 
> resize_reiserfs itself??

Hard to tell actually. Looks like filesystem have the size
changed, but partition table was not adjusted accordingly. Note
that resize_reiserfs does not touch partition tables at all.

> It could also be due to physical problems with the hard drive, which had 
> reported bad blocks some time ago. Is this more likely the cause of the 
> problem??

Well, it does not look like hard drive problem.

Bye,
    Oleg

Re: Can resize_reiserfs do non-destructive resizing??

[Bruce]

Thanks; I ran 
	reiserfsck --rebuild-sb /de/hda2
then tried to run 
	reiserfsck --rebuild-tree
However, running the second came up with:

Pass 0:
####### Pass 0 #######
Loading on-disk bitmap .. ok, 1837081 blocks marked used
Skipping 8373 blocks (super block, journal, bitmaps) 1828708 blocks will be 
read
0%                                                      left 1823494, 2607 
/sec
bread: Cannot read a block # 17599.



Aborted
[root@desktop /] #

Thoughts??

Bruce

Re: Can resize_reiserfs do non-destructive resizing??

[Oleg Drokin]

Hello!

On Mon, Nov 18, 2002 at 08:51:54AM -0500, Bruce wrote:

> Thanks; I ran 
> 	reiserfsck --rebuild-sb /de/hda2
> then tried to run 
> 	reiserfsck --rebuild-tree
> However, running the second came up with:
> Pass 0:
> ####### Pass 0 #######
> Loading on-disk bitmap .. ok, 1837081 blocks marked used
> Skipping 8373 blocks (super block, journal, bitmaps) 1828708 blocks will be 
> read
> 0%                                                      left 1823494, 2607 
> /sec
> bread: Cannot read a block # 17599.

Hm, this is a bad sign.
Try to run badblocks on yor drive.

Is there anything in kernel logs at this time?

Bye,
    Oleg

Re: Can resize_reiserfs do non-destructive resizing??

[Bruce]

I ran "badblocks /dev/hda2" and it produced a list of 39 numbers.

Running the above produced warnings in /var/log/kernel/warnings; the first was 
as follows:

Nov 19 09:52:07 desktop kernel: hda: dma_intr: status=0x51 { DriveReady 
SeekComplete Error }
Nov 19 09:52:07 desktop kernel: hda: dma_intr: error=0x40 { UncorrectableError 
}, LBAsect=140892, sector=140720

And the following in /var/log/kernel/info:

Nov 19 09:52:07 desktop kernel: end_request: I/O error, dev 03:02 (hda), 
sector 140720

The above errors were repeated a number of times with different sectors.

Bruce

On Monday 18 Nov 2002 8:53 am, Oleg Drokin wrote:
> Hello!
>
> On Mon, Nov 18, 2002 at 08:51:54AM -0500, Bruce wrote:
> > Thanks; I ran
> > 	reiserfsck --rebuild-sb /de/hda2
> > then tried to run
> > 	reiserfsck --rebuild-tree
> > However, running the second came up with:
> > Pass 0:
> > ####### Pass 0 #######
> > Loading on-disk bitmap .. ok, 1837081 blocks marked used
> > Skipping 8373 blocks (super block, journal, bitmaps) 1828708 blocks will
> > be read
> > 0%                                                      left 1823494,
> > 2607 /sec
> > bread: Cannot read a block # 17599.
>
> Hm, this is a bad sign.
> Try to run badblocks on yor drive.
>
> Is there anything in kernel logs at this time?
>
> Bye,
>     Oleg

Re: Can resize_reiserfs do non-destructive resizing??

[Oleg Drokin]

Hello!

On Tue, Nov 19, 2002 at 10:48:21AM -0500, Bruce wrote:

> I ran "badblocks /dev/hda2" and it produced a list of 39 numbers.
> Nov 19 09:52:07 desktop kernel: hda: dma_intr: status=0x51 { DriveReady 
> SeekComplete Error }
Well, that means that your hard drive is failing and reiserfs cannot read some
of the data because of bad sectors.

Bye,
    Oleg

Re: Can resize_reiserfs do non-destructive resizing??

[Bruce]

Excellent!

Seriously, I thought as much; I think it is still under warranty, and I do 
have a backup, so no great loss.

Thanks for the feedback.

B.

On Tuesday 19 Nov 2002 10:56 am, Oleg Drokin wrote:
> Hello!
>
> On Tue, Nov 19, 2002 at 10:48:21AM -0500, Bruce wrote:
> > I ran "badblocks /dev/hda2" and it produced a list of 39 numbers.
> > Nov 19 09:52:07 desktop kernel: hda: dma_intr: status=0x51 { DriveReady
> > SeekComplete Error }
>
> Well, that means that your hard drive is failing and reiserfs cannot read
> some of the data because of bad sectors.
>
> Bye,
>     Oleg

Re: Can resize_reiserfs do non-destructive resizing??

[Dieter Nützel]

Am Dienstag, 19. November 2002 16:56 schrieb Oleg Drokin:
> Hello!
>
> On Tue, Nov 19, 2002 at 10:48:21AM -0500, Bruce wrote:
> > I ran "badblocks /dev/hda2" and it produced a list of 39 numbers.
> > Nov 19 09:52:07 desktop kernel: hda: dma_intr: status=0x51 { DriveReady
> > SeekComplete Error }
>
> Well, that means that your hard drive is failing and reiserfs cannot read
> some of the data because of bad sectors.

Yah, replace that disk (or disks).
Some harddisk manufactures like IBM for example tell you to try there DFT tool 
first to get a valid RMA number or "low level" reformat the disk, but we 
replace all such costumer disks. They are out of reserved sector => bad.

Regards,
	Dieter
-- 
Dieter Nützel
Graduate Student, Computer Science

University of Hamburg
Department of Computer Science
@home: Dieter.Nuetzel at hamburg.de (replace at with @)

Re: Can resize_reiserfs do non-destructive resizing??

[Todd Lyons]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dieter Nützel wanted us to know:

>Some harddisk manufactures like IBM for example tell you to try there DFT tool 
>first to get a valid RMA number or "low level" reformat the disk, but we 

Very familiar with those.  We got a stack of boxen with the IBM
DTLA-307030 drives in them.  Lose about 2 per month.

>replace all such costumer disks. They are out of reserved sector => bad.

Netware used to call this "hotfix".  Is there a common name for when the
harddrives do this themselves?
- -- 
Blue skies...		Todd
| Get a bigger hammer!   |  ...file transfer by email is evil.    |
| http://www.mrball.net  |                              NANOG     |
| http://faq.mrball.net  |                                        |
   Linux kernel 2.5.47-1   1 user,  load average: 0.07, 0.02, 0.00
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE92mf2IBT1264ScBURAooGAKCZ2EH3lmVmBYqYZ5W8aRxK6guHDgCgkmya
k+VRFwk8ZjtJgwdQz8qJfgY=
=QjIj
-----END PGP SIGNATURE-----

Re: Can resize_reiserfs do non-destructive resizing??

[Dieter Nützel]

Am Dienstag, 19. November 2002 17:33 schrieb Todd Lyons:
> Dieter Nützel wanted us to know:
> >Some harddisk manufactures like IBM for example tell you to try there DFT
> > tool first to get a valid RMA number or "low level" reformat the disk,
> > but we
>
> Very familiar with those.  We got a stack of boxen with the IBM
> DTLA-307030 drives in them.  Lose about 2 per month.

Last here from Fujitsu the ATA "gel/liquid layer" ones.

> >replace all such costumer disks. They are out of reserved sector => bad.
>
> Netware used to call this "hotfix".  Is there a common name for when the
> harddrives do this themselves?

sector remapping?

Or do you mean the whole reformat? The last could only be initiated through 
the admin and need good backup (dd_rescue) first.

-Dieter

-- 
Dieter Nützel
Graduate Student, Computer Science

University of Hamburg
Department of Computer Science
@home: Dieter.Nuetzel at hamburg.de (replace at with @)