Re: ReiserFS 3.6.x vs Reiser4

[Hans Reiser <reiser@namesys.com>]

Nikita Danilov wrote:

>Oleg O. Ossovitskii writes:
> > Hello, All!
> > 
> > I made simple FS comparision. From a user point of view - unpacking
> > files, copying, deleting etc. Reiser4 is more space effectively, and
> > it remove files faster than reiserfs3, but speed of copying many small files
> > (kernel 2.5.44 source code) is approximately equal to reiserfs3.6. But
> > of reiser4 reliability is still low.
> > 
> > root@(vc/1)[/mnt/reiser4tst]# fdisk -l /dev/hda
> > 
> > Disk /dev/hda: 255 heads, 63 sectors, 4982 cylinders
> > Units = cylinders of 16065 * 512 bytes
> > 
> >    Device Boot    Start       End    Blocks   Id  System
> >    /dev/hda1   *         1        66    530113+   6  FAT16
> >    /dev/hda2            67        83    136552+  82  Linux swap
> >    /dev/hda3            84        86     24097+  83  Linux
> >    /dev/hda4            87      4981  39319087+   f  Win95 Ext'd (LBA)
> >    /dev/hda5            87       405   2562336    7  HPFS/NTFS
> >    /dev/hda6           406      1425   8193118+   7  HPFS/NTFS
> >    /dev/hda7          1426      2190   6144831   83  Linux
> >    /dev/hda8          2191      2321   1052226   83  Linux
> >    /dev/hda9          2322      2582   2096451    b  Win95 FAT32
> >    /dev/hda10         2583      3220   5124703+   b  Win95 FAT32
> >    /dev/hda11         3221      4495  10241406    7  HPFS/NTFS
> >    /dev/hda12         4496      4619    995998+  83  Linux
> >    /dev/hda13         4620      4743    995998+  83  Linux
> > 
> > 
> > /dev/hda12 is Reiser4
> > /dev/hda13 is ReiserFs 3.6.25
> > 
> > root@(vc/1)[/mnt/reiser4tst]# df -ah .
> > Filesystem            Size  Used Avail Use% Mounted on
> > /dev/hda12            973M   29M  944M   3% /mnt/reiser4tst
> > 
> > root@(vc/1)[/mnt/reiser3tst]# df -ah .
> > Filesystem            Size  Used Avail Use% Mounted on
> > /dev/hda13            973M   61M  912M   6% /mnt/reiser3tst
> > 
> > Tests description:
> > 1. unpacking linux-2.5.44.tar.bz2
> >    $ ls -l linux-2.5.44.tar.bz2
> >    29260
> > 2. cp -Rd linux-2.5.44 linux-2.5.44-test1
>
Please create the tarball from a reiser4 directory and repeat the 
benchmark if you could.

Your results are completely inconsistent with Zam's as far as I 
understand them.  Zam, please reproduce and comment.

> > 3. du -sk linux-2.5.44
> > 4. rm -rf linux-2.5.44-tst1
> > 5. rm -rf linux-2.5.44
> > 
> > -------------+--------------------+---------------------
> > Test number  |   ReiserFS 3.6.25  |   Reiser4
> >     1        |       1m22.868s    |   1m24.153s
> >     2        |       1m24.127s    |   1m23.050s
> >     3        |  0m5.727s/183656k  | 0m3.706s/157401k
> >     4        |       0m9.896s     |   0m3.185s(*)
> >     5        |       0m6.921s     |   0m33.011s
> > -------------+--------------------+---------------------
> > (*) rm couldn't remove directories recursively, and gave warnings/errors:
> >     rm: cannot remove directory `linux-2.5.44-tst1/arch/um': Directory not empty
> >     rm: cannot remove directory `linux-2.5.44-tst1/arch/x86_64': Directory not empty
> >     ... and so on 
>
>Are there any messages in the kernel log at this time? Can you reproduce
>this with CONFIG_REISER4_CHECK/Assertions on?
>
> >     After that Reiser4 partition stay unmountable and have unremovable 
> >     directories till reboot
> >     After reboot directories was succesfully removed, but they still contained
> >     files
>
>You mean, rm still complains about not empty directories? Can you please
>check whether they actually *are*?
>
>It would be great if you could reproduce it and send to me
>(personally---too large for list) strace of failing rm.
>
> >     Test #5 done after reboot - I got significant Reiser4 slowdown
> > 
> > --- Sys info ---
> > # uname -a
>
>Nikita.
>
>
>  
>