0

I have a reiserfs volume that required a --rebuild-tree, but is currently failing to complete when I pass it --rebuild-tree. Here is the output that I receive when running it:

reiserfsck 3.6.19 (2003 www.namesys.com)


###########
reiserfsck --rebuild-tree started at Mon Oct 26 13:22:16 2009
###########

Pass 0:
####### Pass 0 #######
The whole partition (7864320 blocks) is to be scanned
Skipping 8450 blocks (super block, journal, bitmaps) 7855870 blocks will be read
0%....20%....40%....60%....80%....100%                        left 0, 9408 /sec
287884 directory entries were hashed with "r5" hash.
    "r5" hash is selected
Flushing..finished
    Read blocks (but not data blocks) 7855870
        Leaves among those 6105606
        Objectids found 287892

Pass 1 (will try to insert 6105606 leaves):
####### Pass 1 #######
Looking for allocable blocks .. finished
0%....20%....40%....60%....80%....Not enough allocable blocks, checking bitmap...there are 1 allocable blocks, btw

out of disk space
Aborted

I can't mount it, and I can't fsck it. I've tried extending the volume, but that hasn't helped either.

mgorven
  • 30,036
  • 7
  • 76
  • 121

3 Answers3

2

Yes, I am using the latest reiserfstools. I had previously tried the dd method to copy the volume to another raid set, but that was failing as well. We extended the volume a second time (so it went from 30GB -> 40GB -> 50GB), and for whatever reason the fsck completed when run on the 50GB version of the volume. Seems odd that it would need to be extended by 20GB to complete, but we're not complaining! Volume then could successfully be mounted and the data copied off.

0

Try to copy the whole Partition to another bigger Disc. (e.g. with dd) Then do a rebuild tree there. I've done a lots of rescues with reiserfs, but this error seems to be really bad. Maybe something is really broken with your disk or its a bug. Are you using the latest resiserfstools?

evildead
  • 892
  • 5
  • 11
  • If you copy the partition to a new disk, don't you have to extend the partion and resize filesystem to make reisefs aware of the new space? Is this a good idea on a broken fs? – Wienczny Oct 27 '09 at 01:38
  • why should this in any way harm the broken filesystem? You mount this fs in read only an copy it with dd to another harddrive. Than you have an imagefile. (maybe I forgot to say that exactly) Now you only operate on that file, an you can work with it. You can expand it eg. or just try to repair it. I think his HD is broken in any way, and so operating only on the real Data will give him the oppurtunity to get the files back. Maybe :) If all goes well, he can loop mount the repaired partition and copy his files to a ne Harddrive. – evildead Oct 27 '09 at 10:54
0

thanks mojo! The same magic saved my data - i got the "no space left" error after --rebuild-tree too, i decided to try upsizing logical volume from 4G to 5G, but it didnt worke either. After reading that mojo was saved by upsizing it again, i gave it one more chance... now from 5G to 10G and it worked!!! I'm Not sure if it wanted more space, or two resizigs do the trick, but i'm glad i have my data back :]