Re: libparted trashes reiserfs and enjoys it! (fwd)

[Szakacsits Szabolcs]

---------- Forwarded message ----------
Date: Fri, 25 Nov 2005 15:46:38 +0200
From: Yuriy Umanets <address@hidden>
To: Szakacsits Szabolcs <address@hidden>
Subject: Re: libparted trashes reiserfs and enjoys it!

Szakacsits Szabolcs wrote:
> Ooops, I've sent it to the wrong address

Hi!

In fact you sent it to correct address. I added reiserfs support to 
parted some time ago ;)

>>>Please don't use progreiserfs/libparted even though 
>>>http://www.gnu.org/software/parted/ says it comes from namesys.com
>>>
>>>I've just been talking to Hans Reiser and all after loosing a load of 
>>>data using qtparted who pointed out that it is NOT written by them and 
>>>is buggy as hell. (Enough to lose my data anyway). They are pulling it 
>>>off their ftp site to avoid misleading people, but can I respectifully 
>>>suggest you shell out to resize_reiserfs to do the dirty work of resizing.
>>
>>Well, reiserfs has been known for eating filesystems for breakfast, as long as
>>i can remember, just don't use it.

Long ago, may be a year or so, there were complains that progsreiserfs 
eats FSs on resize, but those ppl. as well did not give me any details 
of problem, so I was not able to fix it.

Also, I've found, that all problems were experienced with only using 
qtparted (and others seem work correctly) on some particular 
distribution (seems gentu, do not remember exactly). And as there were 
cases in past, that distomakers modified reiserfsprogs and after that 
fsck for instance did not work well, I stopped to dig that issue.

I still want to add, if somebody gives me a prove and/or details of any 
bugs in libreiserfs, I'll fix it asap. Or let people who knows the 
source of issue fix it them selves, i do not mind ;)

I suspect that there may be bugs in resize code. Resize code is quite 
simple, but those times I added it we had no such a big HDD to test it 
on ;)

> 
> 
> There wasn't anything in the email indicating reiserfs problem: zero
> information like version numbers, error messages, old and new partition
> table, etc.

> 
> It's a well known fact that QTParted sometimes needlessly shifts 
> the partition start (when libparted uses the wrong geometry) and 
> as a consequence the filesystem becomes inaccessible.

I did not know that.

  This is
> always interpreted by the users as the fault of the filesystem 
> resizer which is obviosly not true and things could be trivially 
> restored.
> 
> 
>>Now, code is welcome to use reiserfsprog library and reimplement the reiserfs
>>resizing support in libparted, but nobody seems to care enough about reiserfs
>>to actually provide code.

That is possible in principle ;) And even could better, as reiserfsprogs 
is supported by Namesys.

However, person saying that, seem knows nothing about reiserfsprogs. 
There is no magical library functions like resize, create, etc. All the 
stuff is located in progs itself and using just library is impossible.
Library there is just a pile of code which is impossiblke to be re-used. 
That was main reason why I wroite libreiserfs.

The better way is to fix bug (if any) in libreiserfs.

As for original reiserfs code providing, what is problem? Download 
reiserfsprogs from namesys ftp and go ahead to use/rework it ;)

> 
> 
> Yury, the maintainer, says no problem in the latest version and before
> making bogus conslusions, it would be nice to know what version users use
> and what is actually the problem.
> 
>       Szaka

Thanks.


-- 
umka