Samba + POSIX ACL, e2fsck

[Jan Derfinak]

On Mon, 24 Nov 2003, Petr Novák wrote:

> Ještě poznámka ke XFS:
> Začal bych ho používat i doporučovat např. kdyby byl rychlejší.
> Takže test - kopie adresáře obsahující cca 3000 podadresářů.
> výsledky:
> XFS   ... 65 sec
> EXT3  ...33 sec
> -> XFS je 2x pomalejší než EXT3.
> Děkuji nechci.

Zaujimave. Mohli by Ste uviest verziu XFS, jadra, velkost a pocet bufrov pri
montovani xfs (ak Ste ich pouzily), typ udajov (boli to len adresare,
alebo tam boli aj subory)?

Napriklad takto:
# mount /dev/lvm2/data /mnt/mnt2
# cd /mnt/mnt2
# dd if=/dev/zero of=disk bs=1M count=512
512+0 records in
512+0 records out
# losetup /dev/loop0 disk
# cat > test
#!/bin/tcsh

@ i = 0
while ( $i < 3000 )
  mkdir $i
  dd if=/dev/zero of=$i/$i bs=1k count=20 >& /dev/null
  @ i++
end
# chmod +x test
# mkfs -t ext3 /dev/loop0 
mke2fs 1.34 (25-Jul-2003)
Filesystem label=
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
131072 inodes, 524288 blocks
26214 blocks (5.00%) reserved for the super user
First data block=1
64 block groups
8192 blocks per group, 8192 fragments per group
2048 inodes per group
Superblock backups stored on blocks: 
        8193, 24577, 40961, 57345, 73729, 204801, 221185, 401409

Writing inode tables: done                            
Creating journal (8192 blocks): done
Writing superblocks and filesystem accounting information: done

This filesystem will be automatically checked every 23 mounts or
180 days, whichever comes first.  Use tune2fs -c or -i to override.
# mount /dev/loop0 /mnt/mnt3
# cd /mnt/mnt3
# mkdir src dst
# cd src
# time /mnt/mnt2/test 

real    0m14.796s
user    0m4.204s
sys     0m7.719s
# time cp -a * ../dst/

real    0m22.333s
user    0m0.108s
sys     0m2.096s
# time rm -rf ../dst/*

real    0m0.511s
user    0m0.035s
sys     0m0.443s
# cd
# umount /mnt/mnt3
# mkfs -t xfs -f /dev/loop0 
meta-data=/dev/loop0             isize=256    agcount=8, agsize=16384 blks
         =                       sectsz=512  
data     =                       bsize=4096   blocks=131072, imaxpct=25
         =                       sunit=0      swidth=0 blks, unwritten=1
naming   =version 2              bsize=4096  
log      =internal log           bsize=4096   blocks=1200, version=1
         =                       sectsz=512   sunit=0 blks
realtime =none                   extsz=65536  blocks=0, rtextents=0
# mount /dev/loop0 /mnt/mnt3
# cd /mnt/mnt3
# mkdir src dst
# cd src
# time /mnt/mnt2/test

real    0m11.587s
user    0m4.033s
sys     0m7.260s
# time cp -a * ../dst/

real    0m1.538s
user    0m0.145s
sys     0m1.204s
# time rm -rf ../dst/*

real    0m0.639s
user    0m0.039s
sys     0m0.500s
# uname -a
Linux test 2.4.21-99-athlon #1 Wed Sep 24 13:34:32 UTC 2003 i686 athlon i386 GNU/Linux
# free
             total       used       free     shared    buffers     cached
Mem:        514764     506268       8496          0         36     433512
-/+ buffers/cache:      72720     442044
Swap:      1052248          0    1052248
# dmesg | grep XFS
SGI XFS 1.3.0 with ACLs, no debug enabled
SGI XFS Quota Management subsystem
# dmesg | grep -i EXT3
EXT3 FS 2.4-0.9.19, 19 August 2002 on loop(7,0), internal journal

Ziadam, aby Ste brali tento test iba ako jednoucelovy priklad, v ziadnom
pripade nemal dokazat, ze xfs je 12x rychlesie pri kopirovani ako ext3.
Skutocne testovanie FS dost komplexna a komplikovana zalezitost.

				jano

--