Re: Reiserfsck dies

["Markus Müller" <mm@priv.de>]

Hi Jan Engelhardt,
>> Hi Linux kernel users,
>>
>> reiserfsck told me that I have to run --rebuild-tree to fix all errors. But
>> this don't work (see below), I tried two times (every time I am waiting 28
>> hours). If I mount the filesystem, there are no files in it. What can I do?
>>     
>
> Is the harddisk broken? (Check for spurious IDE warnings in /var/log/...)
> It is possible that fsck zeroed out many entries because they were not 
> readable.
>   
no, the hdd is a software raid 5, /dev/md0, which is piped through aes
via cryptsetup, so it is accessed via /dev/mapper/hdb (it is not
/dev/hdb, which you could think about the name, but /dev/md0). There
were memory problems, so the reiserfs got inconsistent. But it STILL
worked before I run reiserfsck --rebuild-tree! Just reiserfsck --check
told me to run this command, this is the only cause I run it.
>   
>> stacker:/# dmesg
>> oop0: warning: vs-5150: search_by_key: invalid format found in block 1770409.
>> Fsck?
>> ReiserFS: loop0: warning: vs-13070: reiserfs_read_locked_inode: i/o failure
>>     
> [...]
>
> reiserfs warnings, but no lowlevel warnings :-/
> Did, by chance, you use dm-flakey?
>   
Please tell me, what is dm-flakey?

Regards
Markus Mueller