Long rm times of ccreg40 R4 files

Long rm times of ccreg40 R4 files

[Dushan Tcholich]

I was restructuring my partition layout on some disk where I had plain
R4, and while doing it I deleted some large files.
I was suprised how short delete times were.
So on cc partition I created incompressable file ( as per Edwards
instructions http://marc.info/?l=reiserfs-devel&m=121262617103900&w=2)
and:

localhost ~ # ls -l /w*
-rw-r--r-- 1 root root 2600005632 Apr 23 10:22 /windows.qcow2
localhost ~ # time rm /windows.qcow2

real    0m0.537s
user    0m0.000s
sys     0m0.456s

But when that file is managed with ccreg40 plugin on same partition:
localhost ~ # ls -l /w*
-r--r--r-- 1 root root 2600005632 Apr 23 10:29 /w.qcow2
localhost ~ # time rm /w.qcow2

real    1m1.523s
user    0m0.000s
sys     0m17.685s

iotop: 2044 be/4 root       16.23 M/s   16.25 M/s  ?unavailable?  rm /w.qcow2

If I can give any more data feel free to ask

Have a nice day

Dushan

PS.
localhost ~ # uname -a
Linux localhost 2.6.31-hh #1 SMP Sat Feb 6 17:58:52 CET 2010 x86_64
Intel(R) Core(TM)2 CPU 6400 @ 2.13GHz GenuineIntel GNU/Linux

/dev/md2                /               reiser4         noatime         0 0

Partition on Raid1 created with plain mkfs.reiser4 -o create=ccreg40

Re: Long rm times of ccreg40 R4 files

[Bartosz Szreder]

I confirm. It takes quite some time to delete anything bigger than 100 MiB.

cfiend@Ebola ~ $ uname -a
Linux Ebola 2.6.33-tuxonice-r1 #1 SMP Tue Apr 13 09:39:51 CEST 2010
x86_64 Pentium(R) Dual-Core CPU T4200 @ 2.00GHz GenuineIntel GNU/Linux

/dev/sda2               /               reiser4         noatime,nodiratime 0 1

Re: Long rm times of ccreg40 R4 files

[Edward Shishkin]

Dushan Tcholich wrote:
> I was restructuring my partition layout on some disk where I had plain
> R4, and while doing it I deleted some large files.
> I was suprised how short delete times were.
> So on cc partition I created incompressable file ( as per Edwards
> instructions http://marc.info/?l=reiserfs-devel&m=121262617103900&w=2)
> and:
>
> localhost ~ # ls -l /w*
> -rw-r--r-- 1 root root 2600005632 Apr 23 10:22 /windows.qcow2
> localhost ~ # time rm /windows.qcow2
>
> real    0m0.537s
> user    0m0.000s
> sys     0m0.456s
>
> But when that file is managed with ccreg40 plugin on same partition:
> localhost ~ # ls -l /w*
> -r--r--r-- 1 root root 2600005632 Apr 23 10:29 /w.qcow2
> localhost ~ # time rm /w.qcow2
>
> real    1m1.523s
> user    0m0.000s
> sys     0m17.685s
>
>   

Data of a large ccreg40 file is scattered on many-many formatted
nodes, so in order to delete such file we need to scan the tree and
perform expensive surgery.

While a reg40 file is just one extent pointer in a formatted node :)

Edward.

> iotop: 2044 be/4 root       16.23 M/s   16.25 M/s  ?unavailable?  rm /w.qcow2
>
> If I can give any more data feel free to ask
>
> Have a nice day
>
> Dushan
>
> PS.
> localhost ~ # uname -a
> Linux localhost 2.6.31-hh #1 SMP Sat Feb 6 17:58:52 CET 2010 x86_64
> Intel(R) Core(TM)2 CPU 6400 @ 2.13GHz GenuineIntel GNU/Linux
>
> /dev/md2                /               reiser4         noatime         0 0
>
> Partition on Raid1 created with plain mkfs.reiser4 -o create=ccreg40
> --
> To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>