reiserfs data recovery

reiserfs data recovery

[D. Sen]

After a hard disk crash, I dont see the reiserfs filesystems anymore 
(cant mount them) although the partitions are there and the ext2 
filesystems are there. What are my best options trying to retrieve any 
data that were there in the reiserfs partions?

DS

Re: reiserfs data recovery

[Manuel Krause]

On 07/28/03 00:39, dsen@homemail.com wrote:
> After a hard disk crash, I dont see the reiserfs filesystems anymore 
> (cant mount them) although the partitions are there and the ext2 
> filesystems are there. What are my best options trying to retrieve any 
> data that were there in the reiserfs partions?
> 
> DS
> 

Try the following from a spare/rescue partition:

A.1) Try the most recent reiserfsprogs from
       ftp://ftp.namesys.com/pub/reiserfsprogs
      on your lost partitions and provide the info of
      "reiserfsck --check /lost/partition"

A.2) Try "reiserfsck --rebuild-tree /lost/partition on your
      own risk (!!!)

B) Use most recent plain kernel (> 2.4.19)

C) Provide more info in general (versions of kernel,
    reiserfsprogs, distro, credit card info ;-)
    and so on...)

D) Refer to http://www.namesys.com/support.html


You can mix the things above, but if you can, use at least the row A.1, 
C, B, A.2, always considering D before the next full report!

Best wishes,

  Manuel

Re: reiserfs data recovery

[D. Sen]

Thanks for that information. When I couldnt mount my reiserfs partition, 
I did do a mkreiserfs on that partition. How is that likely to affect 
your suggestion A.2?

I was using kernel 2.4.20, Mandrake 8.1...and the reiserfs programs that 
came with it.

I used the reiserfsck that came with the Mandrake 9.1 distro...to make 
the filesystem.

Credit Card information to follow...

*sigh*

Manuel Krause wrote:
> On 07/28/03 00:39, dsen@homemail.com wrote:
> 
>> After a hard disk crash, I dont see the reiserfs filesystems anymore 
>> (cant mount them) although the partitions are there and the ext2 
>> filesystems are there. What are my best options trying to retrieve any 
>> data that were there in the reiserfs partions?
>>
>> DS
>>
> 
> Try the following from a spare/rescue partition:
> 
> A.1) Try the most recent reiserfsprogs from
>       ftp://ftp.namesys.com/pub/reiserfsprogs
>      on your lost partitions and provide the info of
>      "reiserfsck --check /lost/partition"
> 
> A.2) Try "reiserfsck --rebuild-tree /lost/partition on your
>      own risk (!!!)
> 
> B) Use most recent plain kernel (> 2.4.19)
> 
> C) Provide more info in general (versions of kernel,
>    reiserfsprogs, distro, credit card info ;-)
>    and so on...)
> 
> D) Refer to http://www.namesys.com/support.html
> 
> 
> You can mix the things above, but if you can, use at least the row A.1, 
> C, B, A.2, always considering D before the next full report!
> 
> Best wishes,
> 
>  Manuel
> 
> 
> 

Reiserfs Data Recovery

[Nick Vahalik]

Hello,

System: Slackware 9.1
Dual P3 1Ghz/133Mhz FSB
1 GB Ram PC133 Ram
Running Hardware Raid1 on top of a highpoint 37x2 raid controller
2x80 GB Drives

My system crashed yesterday (xscreensaver... bah!) and when I rebooted 
my /raid partition wouldn't mount. So I ran reiserfsck /dev/sda1 and it 
said I needed to rebuild my superblock because my Root Super Block 0 was 
corrupt. So I ran reiserfsck --rebuild-sb /dev/sda1 and after that it 
was saying that it failed and I needed to run reiserfsck --rebuild-tree. 
Well, I did and about an hour later I get something like this (the log 
is about 80k):
1537 directory entries were hashed with not set hash.
112120 directory entries were hashed with "r5" hash.
Selected hash ("r5") does not match to the hash set in the super block 
(not set).
"r5" hash is selected
Flushing..finished
Read blocks (but not data blocks) 38667615
Leaves among those 56626
- corrected leaves 2478
- leaves all contents of which could not be saved and deleted 19
pointers in indirect items to wrong area 5167 (zeroed)
Objectids found 116706

Pass 1 (will try to insert 56607 leaves):
####### Pass 1 #######
Looking for allocable blocks .. finished
0%pass1: block 65604, item 16, entry 4: The entry "P313"?" of the [50706 
50722 0x1 DIR (3)] is hashed with not set whereas proper hash is "r5" - 
deleted
pass1: block 65604, item 16, entry 9: The entry "P313"?" of the [50706 
50722 0x1 DIR (3)] is hashed with not set whereas proper hash is "r5" - 
deleted
pass1.c 408 pass1_correct_leaf
pass1_correct_leaf: block 65636, item 0, pointer 7: The wrong pointer 
(2224053807) in the file [50872 50891]. Must be fixed on pass0.
Aborted

So, I figured I'd run it once more and just hope and pray that it would 
kinda fix itself, and well, I got the same thing again:

1200 directory entries were hashed with not set hash.
110404 directory entries were hashed with "r5" hash.
"r5" hash is selected
Flushing..finished
Read blocks (but not data blocks) 38667391
Leaves among those 56589
- corrected leaves 2484
- leaves all contents of which could not be saved and deleted 28
pointers in indirect items to wrong area 4947 (zeroed)
Objectids found 116691

Pass 1 (will try to insert 56561 leaves):
####### Pass 1 #######
Looking for allocable blocks .. finished
0%pass1: block 65540, item 19, entry 0: The entry "" of the [50257 50323 
0x1 DIR (3)] is hashed with not set whereas proper hash is "r5" - deleted
pass1: block 65540, item 19, entry 0: The entry ".." of the [50257 50323 
0x2 DIR (3)] is hashed with not set whereas proper hash is "r5" - deleted
pass1.c 408 pass1_correct_leaf left 56393, 0 /sec
pass1_correct_leaf: block 65604, item 2, pointer 1: The wrong pointer 
(1) in the file [50696 50698]. Must be fixed on pass0.
Aborted

It looks like the numbers are getting smaller, but all I know is that it 
still won't mount... and I have everything on this partition... this was 
my backup (I just reformatted) and without this data I am seriously 
screwed. Will normal data recovery methods work (like, Encase or 
Ontrack?) or is there a better way?

Nick Vahalik

Re: Reiserfs Data Recovery

[Philippe Gramoullé]

Hello,

What is your reiserfsck version ( reiserfsck -V ? )

Thanks,

Philippe

Re: Reiserfs Data Recovery

[Nick Vahalik]

Philippe,

root@bardiel:~# reiserfsck -V
reiserfsck 3.6.11 (2003 www.namesys.com)

Nick Vahalik

Philippe Gramoulle' wrote:

>Hello,
>
>What is your reiserfsck version ( reiserfsck -V ? )
>
>Thanks,
>
>Philippe
>
>
>  
>

Re: Reiserfs Data Recovery

[Philippe Gramoullé]

Hello,

I'd suggest that you try first the latest reiserfsprogs 3.6.13.

ftp.namesys.com:/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz

It problem still occurs, then someone at Namesys might help you more than me.

Thanks,

Philippe



On Mon, 08 Mar 2004 08:43:56 -0600
Nick Vahalik <nick@nsanity.com> wrote:

  | Philippe,
  | 
  | root@bardiel:~# reiserfsck -V
  | reiserfsck 3.6.11 (2003 www.namesys.com)
  | 
  | Nick Vahalik
  | 

Re: Reiserfs Data Recovery

[Nick Vahalik]

Hello again,

Okay, I updated reiserfsprogs and this is what I got:

0%....20%....40%....60%....80%....100%                       left 0, 
15217 /sec
        "r5" hash is selected
Flushing..finished
        Read blocks (but not data blocks) 38667613
                Leaves among those 56652
                        - corrected leaves 2500
                        - leaves all contents of which could not be 
saved and de leted 26
                pointers in indirect items to wrong area 5332 (zeroed)
                Objectids found 116644
 
Pass 1 (will try to insert 56626 leaves):
Looking for allocable blocks .. finished
0%pass1.c 408 pass1_correct_leaf                             left 56623, 
3 /sec
pass1_correct_leaf: block 65604, item 0, pointer 1: The wrong pointer 
(379082642 5) in the file [50696 50697]. Must be fixed on pass0.
Aborted

I can post the logfile... if anyone wants to see it.

Nick Vahalik

Philippe Gramoullé wrote:

>Hello,
>
>I'd suggest that you try first the latest reiserfsprogs 3.6.13.
>
>ftp.namesys.com:/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz
>
>It problem still occurs, then someone at Namesys might help you more than me.
>
>Thanks,
>
>Philippe
>
>
>
>On Mon, 08 Mar 2004 08:43:56 -0600
>Nick Vahalik <nick@nsanity.com> wrote:
>
>  | Philippe,
>  | 
>  | root@bardiel:~# reiserfsck -V
>  | reiserfsck 3.6.11 (2003 www.namesys.com)
>  | 
>  | Nick Vahalik
>  | 
>
>
>  
>

Re: Reiserfs Data Recovery

[Ryan Underwood]

[-- Attachment #1: Type: text/plain, Size: 1508 bytes --]


On Mon, Mar 08, 2004 at 08:22:49AM -0600, Nick Vahalik wrote:
> My system crashed yesterday (xscreensaver... bah!) and when I rebooted 
> my /raid partition wouldn't mount. So I ran reiserfsck /dev/sda1 and it 
> said I needed to rebuild my superblock because my Root Super Block 0 was 
> corrupt. So I ran reiserfsck --rebuild-sb /dev/sda1 and after that it 
> was saying that it failed and I needed to run reiserfsck --rebuild-tree. 

Note: whatever you do, do _not_ touch the second drive of your array
with reiserfsck.  Assuming it is still in the state where it was before
you started playing around, it may be the _only_ thing that ends up
saving your ass!  When you need a new copy of the corrupt filesystem to
play with, just dd the mirror drive *TO* the playground drive (don't
accidentally confuse the if and of), and only play around with the copy
you made.  If worse comes to worse, having a copy of the filesystem that
is not messed around with, will give the RFS pay-for support a better
chance of getting all your data back.

> screwed. Will normal data recovery methods work (like, Encase or 
> Ontrack?) or is there a better way?

I don't know of any analytic data recovery tool that works with
ReiserFS.  I think your best bet, especially if you depend on this data,
is to suck it up and pay the $20 before something worse happens.
Think about which is worth more... a consulting fee, or all your work.

Good luck!

-- 
Ryan Underwood, <nemesis@icequake.net>

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Reiserfs Data Recovery

[Ryan Underwood]

[-- Attachment #1: Type: text/plain, Size: 2022 bytes --]

On Mon, Mar 08, 2004 at 01:25:30PM -0600, Nick Vahalik wrote:
> >
> >Note: whatever you do, do _not_ touch the second drive of your array
> >with reiserfsck.  Assuming it is still in the state where it was before
> >...
> >
> I thought I mentioned that this was a Raid 0.  There is no mirroring.  
> It's a stripe.
> I know.  Just tell me what I already assumed:  I'm screwed. 

Whoops.  In your initial post:
----------------
Hello,

System: Slackware 9.1
Dual P3 1Ghz/133Mhz FSB
1 GB Ram PC133 Ram
Running Hardware Raid1 on top of a highpoint 37x2 raid controller
2x80 GB Drives
----------------

:-(
Yeah, if it's raid 0, you don't have much of a chance besides giving the
filesystem as it is to someone who knows exactly what they are doing wrt
ReiserFS.

I'm having trouble figuring out how exactly a system crash would have
wiped out your super-block...

> Well, I was thinking more of forensics analysis tools... (Network 
> Security Major at ***********) and we have some neat 
> equipment/software over there... but I'm not sure how much of it will 
> work with ReiserFS...

Without something that understands the logical structures of the
specific filesystem in question (Reiser), it will be difficult to get
anything done.  I can't really offer any help myself because I was in
the same boat you were a few weeks ago.  Fortunately, it was just /usr,
not my life work.  Plus, it was apparently a conspiracy on my part to
destroy my data, judging from the steps I took (wiping out 2 perfectly
good backups in the process of trashing it); it wasn't my system
randomly borking itself like yours seems to have.

I wish I could offer more help, but my advice based on my own experience
is to not do anymore damage than has already been done and to get expert
help.  In the end I could barely retrieve anything from my filesystem
and in the process messed up any chances anyone else would have had to
help me out.  Doh! :)

-- 
Ryan Underwood, <nemesis@icequake.net>

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Reiserfs Data Recovery

[Ryan Underwood]

[-- Attachment #1: Type: text/plain, Size: 2601 bytes --]


On Mon, Mar 08, 2004 at 04:32:11PM -0600, Nick Vahalik wrote:
> Same here.  Of course, now I'm wondering if the crash had anything to do 
> with a possible faulty drive.  It's one of those IBM Deathstars!  I 
> forgot about it until about 2 am this morning when I found my receipt.  
> Eh, well at least I had my code backed up...

Oh wow, you do definitely want to run IBM DFT as soon as possible.  That
might be the only thing that is wrong... usually you would see some
syslog errors though when you tried to mount the drives if the system
was having problems reading from the drive(s).  DFT should not destroy
any data or make anything worse, but it will try to relocate the sector
if it is having trouble reading it.

this firmware upgrade was released a while ago and corrects some
deathstar crashing if you have one of the listed drives:
http://www-1.ibm.com/support/docview.wss?rs=0&q=DTLA&uid=psg1MIGR-44195

> Well, I was thinking about doing more of a raw data recovery.  We have 
> tools that will just simply look for files minus the structure and dump 
> the results out into a spare drive.  I did this once to another web 
> server I had.  It recovered 80% of my code... and the MBR/SB was 
> completely eradicated on that drive as well.

Yeah, I guess you can try something if you have the tools, but just
guessing that to get anything serious done you will need tools that
understand Reiser structures.  Otherwise all you can do in a general
case is to find contiguous allocation blocks and try to piece together
files from those.  That approach might get all of your data, though
pieceing it together would be loads of fun...

> Well you replied, and I guess this is just another lesson to backup, 
> backup, backup!  In all honesty, I don't think the filesystem itself is 
> damaged, I still think it's just the superblock.  I've had this problem 
> before with another drive... it just decided to MBR/Root SB... It was 
> also an Hitachi (IBM) Drive.  *sigh*  Oh well, I'll just set the drives 
> aside and leave em.  Maybe in a week or two I'll have enough money to 
> actually afford $25. 

I think you might try the DFT approach before giving up, but definitely
don't do anymore damage with reiserfsck. :)  Also, since this is a
stripe set, is the md device able to start the raid successfully?  If
so, and if it only fails when you try to mount the filesystem on the
raid, you will have better chances than if md can't even figure out that
you have a stripe set spanning those disks.

-- 
Ryan Underwood, <nemesis@icequake.net>

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Reiserfs Data Recovery

[Vitaly Fertman]

Hello Nick,

On Monday 08 March 2004 19:06, Nick Vahalik wrote:
> Hello again,
>
> Okay, I updated reiserfsprogs and this is what I got:
>
> 0%....20%....40%....60%....80%....100%                       left 0,
> 15217 /sec
>         "r5" hash is selected
> Flushing..finished
>         Read blocks (but not data blocks) 38667613
>                 Leaves among those 56652
>                         - corrected leaves 2500
>                         - leaves all contents of which could not be
> saved and de leted 26
>                 pointers in indirect items to wrong area 5332 (zeroed)
>                 Objectids found 116644
>
> Pass 1 (will try to insert 56626 leaves):
> Looking for allocable blocks .. finished
> 0%pass1.c 408 pass1_correct_leaf                             left 56623,
> 3 /sec
> pass1_correct_leaf: block 65604, item 0, pointer 1: The wrong pointer
> (379082642 5) in the file [50696 50697]. Must be fixed on pass0.
> Aborted

When such things happen especially in different places of the 
tree at every fsck run this often indicates some hardware problem.
I would check the memory with memtest first.

-- 
Thanks,
Vitaly Fertman

Re: Reiserfs Data Recovery

[Hans Reiser]

Ryan Underwood wrote:

>
>Running Hardware Raid1 on top of a highpoint 37x2 raid controller
>2x80 GB Drives
>----------------
>  
>
I had really bad experiences with highpoint.

-- 
Hans

Re: Reiserfs Data Recovery

[Nick Vahalik]

Hans Reiser wrote:

> Ryan Underwood wrote:
>
>>
>> Running Hardware Raid1 on top of a highpoint 37x2 raid controller
>> 2x80 GB Drives
>> ----------------
>>
>>
> I had really bad experiences with highpoint.
>
Actually, I messed up... It should have been raid 0 instead of raid 1. 
But I haven't had any problems so far. I did test the memory of my 
system and memtest didn't give any errors. I will, however run the IBM 
DFT tool and see if it can detect any error with the drives.

Re: Reiserfs Data Recovery

[Hans Reiser]

Nick Vahalik wrote:

> Hans Reiser wrote:
>
>> Ryan Underwood wrote:
>>
>>>
>>> Running Hardware Raid1 on top of a highpoint 37x2 raid controller
>>> 2x80 GB Drives
>>> ----------------
>>>
>>>
>> I had really bad experiences with highpoint.
>>
> Actually, I messed up... It should have been raid 0 instead of raid 1. 
> But I haven't had any problems so far. I did test the memory of my 
> system and memtest didn't give any errors. I will, however run the IBM 
> DFT tool and see if it can detect any error with the drives.
>
>
My bad experience was with getting their drivers to work with 2.6, and I 
tried to use raid 1..., and I figured out that it wasn't really hardware 
raid because it needed a software driver that wasn't in 2.6 and it was 
work to figure out that it was missing and they don't provide telephone 
support and, oh, I just hated them by the end of the experience....  I 
ended up just using each drive as a separate device and their hardware 
as a dumb ide controller.

-- 
Hans

Re: Reiserfs Data Recovery

[Nick Vahalik]

Hans Reiser wrote:

> My bad experience was with getting their drivers to work with 2.6, and 
> I tried to use raid 1..., and I figured out that it wasn't really 
> hardware raid because it needed a software driver that wasn't in 2.6 
> and it was work to figure out that it was missing and they don't 
> provide telephone support and, oh, I just hated them by the end of the 
> experience....  I ended up just using each drive as a separate device 
> and their hardware as a dumb ide controller.
>
Indeed, I too contacted them about a 2.6 driver and a few days later the 
only I response was that work was 'in progress' on a 2.6 driver.  Some 
of their newer hardware is already supported (they said, but I haven't 
checked).  Yeah, next time I'll just buy a real raid controller (and 
hopefully some better drives!).

Nick Vahalik

Re: Reiserfs Data Recovery

[Ryan Underwood]

[-- Attachment #1: Type: text/plain, Size: 1046 bytes --]

On Tue, Mar 09, 2004 at 06:45:18PM +0300, Hans Reiser wrote:
> >
> My bad experience was with getting their drivers to work with 2.6, and I 
> tried to use raid 1..., and I figured out that it wasn't really hardware 
> raid because it needed a software driver that wasn't in 2.6 and it was 
> work to figure out that it was missing and they don't provide telephone 
> support and, oh, I just hated them by the end of the experience....  I 
> ended up just using each drive as a separate device and their hardware 
> as a dumb ide controller.

Yeah, effectively all those cheap "RAID" cards (those w/o a CPU or
local cache) do is provide you with a pretty BIOS menu to setup your
arrays.  All the work is done by the md driver anyway.  The BIOS stores
the array configuration at the end of the disks (I think) and the
"driver" just picks up that configuration and starts running a md set
over the assocated disks.  The hardware itself is a basic ATA controller
as far as I can tell.

-- 
Ryan Underwood, <nemesis@icequake.net>

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Reiserfs Data Recovery

[Nick Vahalik]

Ryan Underwood wrote:

>Oh wow, you do definitely want to run IBM DFT as soon as possible.  That
>might be the only thing that is wrong... usually you would see some
>syslog errors though when you tried to mount the drives if the system
>was having problems reading from the drive(s).  DFT should not destroy
>any data or make anything worse, but it will try to relocate the sector
>if it is having trouble reading it.
>  
>
Well, I ran the DFT, Quick, Advanced, and Bad Sector Tests on all three 
of my drives.  (One boot and the two in the stripe) and they all 
completely successfully without any errors.  The drives appear to have 
checked out.

>I think you might try the DFT approach before giving up, but definitely
>don't do anymore damage with reiserfsck. :)  Also, since this is a
>stripe set, is the md device able to start the raid successfully?  If
>so, and if it only fails when you try to mount the filesystem on the
>raid, you will have better chances than if md can't even figure out that
>you have a stripe set spanning those disks.
>  
>
I'm using highpoint's source code driver (hpt37x2.o) compiled for my SMP 
kernel.
...
But... when I dmesg I only see one processor...? Could that be an 
issue... I'm gonna recompile and try it again.

Nick Vahalik

Re: Reiserfs Data Recovery

[Nick Vahalik]

'lo All,

I've been doing some reading through some of the "hidden features" in 
the source code...

Are these features tested?  Is there any more documentation on them 
besides just what's in the comments?  Are there any good sites to read 
up some more on reiserfs v3.6 internals?  I've become determined to get 
this data back.  I've got a couple ideas that I'd like to try but look 
at some of these commands I see that some of the things I'd like to 
access may already be written.

If the problem is that there are sections of my tree missing, why can't 
we just chop the bad sections off and salvage the data that is still 
there?  Are there any tools that do this?

Nick Vahalik

Re: Reiserfs Data Recovery

[Vladimir Saveliev]

Hello

On Wed, 2004-03-10 at 06:07, Nick Vahalik wrote:
> 'lo All,
> 
> I've been doing some reading through some of the "hidden features" in 
> the source code...
> 
> Are these features tested?  

Most of these things were added quickly to solve certain user problems.
To use them one is to understand what they were created for.

> Is there any more documentation on them 
> besides just what's in the comments?

no.

>   Are there any good sites to read 
> up some more on reiserfs v3.6 internals? 

namesys.com had description of reiserfs internals. But now it does not.
read source code instead

>  I've become determined to get 
> this data back.  I've got a couple ideas that I'd like to try but look 
> at some of these commands I see that some of the things I'd like to 
> access may already be written.
> 
> If the problem is that there are sections of my tree missing, why can't 
> we just chop the bad sections off and salvage the data that is still 
> there?  Are there any tools that do this?
> 

did you try reiserfsck?

> Nick Vahalik
> 

Re: Reiserfs Data Recovery

[Nick Vahalik]

Vladimir Saveliev wrote:

>>Are these features tested?  
>>    
>>
>
>Most of these things were added quickly to solve certain user problems.
>To use them one is to understand what they were created for.
>
>  
>
And I guess that isn't documented....

>>Is there any more documentation on them 
>>besides just what's in the comments?
>>    
>>
>
>no.
>  
>

>>  Are there any good sites to read 
>>up some more on reiserfs v3.6 internals? 
>>    
>>
>
>namesys.com had description of reiserfs internals. But now it does not.
>read source code instead
>  
>
Is there a reason?  I grabbed a copy of the whitepaper from archive.org 
and started reading it last night.  Is that information out of date... 
or is it taken down pending the release of Reiserv4?

>> I've become determined to get 
>>this data back.  I've got a couple ideas that I'd like to try but look 
>>at some of these commands I see that some of the things I'd like to 
>>access may already be written.
>>
>>If the problem is that there are sections of my tree missing, why can't 
>>we just chop the bad sections off and salvage the data that is still 
>>there?  Are there any tools that do this?
>>    
>>
>did you try reiserfsck?
>  
>
Yes... with --rebuild-tree, and it aborts after Pass0 every time.  It 
fixes... maybe 10 leaves... and then it finds a bad pointer (or 
something) and dies every time it runs.

Nick Vahalik

Re: Reiserfs Data Recovery

[Vladimir Saveliev]

Hello

On Wed, 2004-03-10 at 18:08, Nick Vahalik wrote:
> Vladimir Saveliev wrote:
> 
> >>Are these features tested?  
> >>    
> >>
> >
> >Most of these things were added quickly to solve certain user problems.
> >To use them one is to understand what they were created for.
> >
> >  
> >
> And I guess that isn't documented....
> 
> >>Is there any more documentation on them 
> >>besides just what's in the comments?
> >>    
> >>
> >
> >no.
> >  
> >
> 
> >>  Are there any good sites to read 
> >>up some more on reiserfs v3.6 internals? 
> >>    
> >>
> >
> >namesys.com had description of reiserfs internals. But now it does not.
> >read source code instead
> >  
> >
> Is there a reason?  I grabbed a copy of the whitepaper from archive.org 
> and started reading it last night.  Is that information out of date... 
> or is it taken down pending the release of Reiserv4?
> 
I have no idea why it was removed. I will try to find out.

> >> I've become determined to get 
> >>this data back.  I've got a couple ideas that I'd like to try but look 
> >>at some of these commands I see that some of the things I'd like to 
> >>access may already be written.
> >>
> >>If the problem is that there are sections of my tree missing, why can't 
> >>we just chop the bad sections off and salvage the data that is still 
> >>there?  Are there any tools that do this?
> >>    
> >>
> >did you try reiserfsck?
> >  
> >
> Yes... with --rebuild-tree, and it aborts after Pass0 every time.  It 
> fixes... maybe 10 leaves... and then it finds a bad pointer (or 
> something) and dies every time it runs.
> 

Are you using reiserfsck from
ftp://ftp.namesys.com/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz?

If yes - please run
debugreiserfs -p /dev/problemdevice | bzip2 -c > problemdevice.bz2
and make problemdevice.bz2 available for downloading.

It should allow us to fix reiserfsck bug.

If you are using older rieserfsck (reiserfsck -V), please try with the
one of
ftp://ftp.namesys.com/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz

> Nick Vahalik
> 

Re: Reiserfs Data Recovery

[Nick Vahalik]

Vladimir Saveliev wrote:

>Are you using reiserfsck from
>ftp://ftp.namesys.com/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz?
>
>If yes - please run
>debugreiserfs -p /dev/problemdevice | bzip2 -c > problemdevice.bz2
>and make problemdevice.bz2 available for downloading.
>
>It should allow us to fix reiserfsck bug.
>
>If you are using older rieserfsck (reiserfsck -V), please try with the
>one of
>ftp://ftp.namesys.com/pub/reiserfsprogs/reiserfsprogs-3.6.13.tar.gz
>  
>
root@bardiel:~/reiserdata# debugreiserfs -V
debugreiserfs 3.6.13 (2003 www.namesys.com)

root@bardiel:~/reiserdata# debugreiserfs -p /dev/sda1 | bzip2 -c > 
devsda1.bz2
debugreiserfs 3.6.13 (2003 www.namesys.com)
 
Loading on-disk bitmap .. 38677079 bits set - done
super block..ok
bitmaps..(1228).. ok
journal (from 18 to 8210)..ok
Super block, bitmaps, journal - 9422 blocks - done, 38667657 blocks left
0%....20%....40%....60%....80%....100%                       left 0, 
15211 /sec
Packed 66463 blocks:
        compessed 63534
        full blocks 2929
                leaves with broken block head 55
                corrupted leaves 83
                internals 672
                descriptors 0
data packed with ratio 0.09

The meta-data info can be downloaded at 
http://24.162.126.168:8080/devsda1.bz2
~5.3 MB

Thanks

Nick Vahalik

Re: Reiserfs Data Recovery

[Nick Vahalik]

Vitaly Fertman wrote:

>Hello,
>
>  
>
>>The meta-data info can be downloaded at
>>http://24.162.126.168:8080/devsda1.bz2
>>~5.3 MB
>>    
>>
>
>I have tried fsck on these metadata and had no problems.
>Did you compile progs on the same conputer where you 
>run it?
>
>  
>
Yes, I did.

Re: Reiserfs Data Recovery

[Stewart Smith]

[-- Attachment #1: Type: text/plain, Size: 454 bytes --]

On Wed, 2004-03-10 at 22:41, Vladimir Saveliev wrote:
> namesys.com had description of reiserfs internals. But now it does not.
> read source code instead

Is it possible to get all of these documents back and in a nice, central
location? Historical documents can be quite interesting, especially in
trying to understand the design process of modern file systems.


-- 
Stewart Smith (stewart@flamingspork.com)
http://www.flamingspork.com/


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]