vs-4080 & more corruptions

vs-4080 & more corruptions

[Christian]

hello,

a few days ago i got a some vs-4080 errors message on my stderr:

vs-4080: reiserfs_free_block: free_block (0341:736255)[dev:blocknr]: bit 
already cleared
vs-4080: reiserfs_free_block: free_block (0341:745214)[dev:blocknr]: bit 
already cleared
vs-4080: reiserfs_free_block: free_block (0341:720895)[dev:blocknr]: bit 
already cleared
vs-4080: reiserfs_free_block: free_block (0341:712573)[dev:blocknr]: bit 
already cleared
vs-4080: reiserfs_free_block: free_block (0341:786297)[dev:blocknr]: bit 
already cleared
vs-4080: reiserfs_free_block: free_block (0341:737149)[dev:blocknr]: bit 
already cleared

i searched the web/reiserfs list for this error and came to the 
cunclusion, that i had to reiserfsck the partition.

(it is a 4,2 GB partition, full-sized for this harddisk. i moved some 
data to/from this partition to another (?) when i got these errors.)

first i did a reiserfsck --check /dev/hdb1, i checked and gave me a log 
file with entries like this:

bad_indirect_item: block 8308: item 3 79 0x5654001 IND (1), len 4048, 
location 48 entry count 0, fsck need 0, format new has a pointer 903 to 
the block 196095 which is in tree already

there are about 20lines like this in the logfile, i cn give you more if 
you want.

but the check finished with tellng me "not all errors could be fixed, 
please run '--rebuild-tree'"

so i did the --rebuild-tree. it replayed some more transactions and then 
continued with "Pass 0":

------
Pass 0:
Loading on-disk bitmap .. ok, 924386 blocks marked used
Skipping 8242 blocks (super block, journal, bitmaps) 916144 blocks will 
be read
0%....20%....40%....60%....80%....100%                        left 0, 
1696 /sec
	"r5" hash is selected
Flushing..done
	Read blocks (but not data blocks) 916144
		Leaves among those 1109
		Objectids found 1618

Pass 1 (will try to insert 1109 leaves):
Looking for allocable blocks .. ok
0%                                                            left 1079, 
0 /sec
mark_block_used: (130815) used already


Aborted.
-------

--> Aborted !? so for now i can't mount the partition:

root@sheep:~/trash# mount -v /dev/hdb1 /mnt/disk/
mount: you didn't specify a filesystem type for /dev/hdb1
        I will try type reiserfs
mount: Not a directory

huh?

i have the logs of all my fsck runs available, i could send it to the 
list if you want to.

i am using a debian/woody system (i386), reiserfsprogs 3.x.1b from the 
woody distribution. --> should i try with other/newer progs?

please help me out of this problem, the data was not so important, but: 
i've had more hdd related / fs-related problems during the last year, 
and always after reiserfsck issues. i really want to hold reiserfs the 
faith, because i think it is a good thing to have a performant fs with 
journal too, but i can't stand the problems anymore.

thanks,
Christian.

Re: vs-4080 & more corruptions

[Oleg Drokin]

Hello!

On Wed, Jun 26, 2002 at 01:33:28AM +0200, Christian wrote:

> Pass 1 (will try to insert 1109 leaves):
> Looking for allocable blocks .. ok
> 0%                                                            left 1079, 
> 0 /sec
> mark_block_used: (130815) used already
> Aborted.
> -------
> --> Aborted !? so for now i can't mount the partition:

So far we came to conclusion that this abort is a clear sign of bad memory or
something damaging data that is coming out of your HDD.
It is easily verifiable. Just run reiserfsck --rebuild-tree again
if it progresses to next passes (means you are lucky and nothing
was damaged this time) or fails with different blocknumber "being in use",
you will know that you need to replace some of your hardware.
(may be something as simple as replacing IDE cable).


> root@sheep:~/trash# mount -v /dev/hdb1 /mnt/disk/
> mount: you didn't specify a filesystem type for /dev/hdb1
>        I will try type reiserfs
> mount: Not a directory
> huh?

unfinished rebuild tree does this, nothing to worry. Once reiserfsck
--rebuild-tree will correctly complete al its operations you will be able to
mount it again.

> please help me out of this problem, the data was not so important, but: 
> i've had more hdd related / fs-related problems during the last year, 
> and always after reiserfsck issues. i really want to hold reiserfs the 
> faith, because i think it is a good thing to have a performant fs with 
> journal too, but i can't stand the problems anymore.

It is impossible to have bad hardware and high stability at the same time
at this level of technology.

Bye,
    Oleg

Re: vs-4080 & more corruptions

[Christian]

> > mark_block_used: (130815) used already
> > Aborted.
> > -------
> > --> Aborted !? so for now i can't mount the partition:
>
> So far we came to conclusion that this abort is a clear sign of bad memory or
> something damaging data that is coming out of your HDD.
> It is easily verifiable. Just run reiserfsck --rebuild-tree again
> if it progresses to next passes (means you are lucky and nothing
> was damaged this time) or fails with different blocknumber "being in use",
> you will know that you need to replace some of your hardware.
> (may be something as simple as replacing IDE cable).

yes, this time it was

"mark_block_usesd: (131069) used already"

but the run aborted again after this. but i don't think that it is bad
memory, because i raun this memory for 1 year now, and problems while
compiling things or running for days at high load or something like that.
always stable. but i will change the ide-cable and place the disk in a
different machine and i will tell you my progress.

> It is impossible to have bad hardware and high stability at the same time
> at this level of technology.

yes, of course, you are right. but i would prefer a failing network card
or a failing power supply again but a failing harddisk/fs is real a bad
thing. and i don't know why always the reiserfs disk are failing. my whole
rootfs is on ext2 (was too lazy to reformat), and no errors in there.
hm, i hope it's the ide-cable.

Thank you very much,
Christian.

-- 
c_kujau@web.de
evilninja@gmx.net

Re: vs-4080 & more corruptions

[Christian]

hi,

> you will know that you need to replace some of your hardware.
> (may be something as simple as replacing IDE cable).

well, you were right. and i was wrong to blame reiserfs for this problem.
i'm very sorry about my calumny.

i set the disk into another system and the --rebuild-tree finished and the
--check did not find any errors anymore and now i can mount and read/write
o my partition and all my data is in there --> i'm very happy.
i changed the cable in this machine and i hope it was the cable and not
the ide-chipset of this quite old machine the disk is now in again.

thank you very much again for your hint and sorry again for bothering this
list with hardware related problems. and thank you for maintaining
reiserfs & this mailing-list.

thanks,
Christian.

-- 
c_kujau@web.de
evilninja@gmx.net

Re: vs-4080 & more corruptions

[Kuba Ober]

On ¶roda 26 czerwiec 2002 09:18 am, Christian wrote:
> hi,
>
> > you will know that you need to replace some of your hardware.
> > (may be something as simple as replacing IDE cable).
>
> well, you were right. and i was wrong to blame reiserfs for this problem.
> i'm very sorry about my calumny.
>
> i set the disk into another system and the --rebuild-tree finished and the
> --check did not find any errors anymore and now i can mount and read/write
> o my partition and all my data is in there --> i'm very happy.
> i changed the cable in this machine and i hope it was the cable and not
> the ide-chipset of this quite old machine the disk is now in again.
>

That's one of the reasons to use newer machines with UDMA drives & 
controllers, because these CRC-check the data going over the wires, which 
helps a lot in some cases.

Cheers, Kuba