OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: Re: high capacity configuration nightmares
From: Matthias Andree (madt.e-technik.uni-dortmund.de)
Date: Sat Oct 28 2000 - 16:42:12 CDT


Dan Hollis <goemonanime.net> writes:

> On Fri, 27 Oct 2000, Wietse Venema wrote:
> > It won't be currupted, but your files will end up in lost+found.
> >
> > Just run the Postfix fsstone utility
> >
> > ./fsstone -cr 10000 10000
> >
> > and hit RESET in the middle of a run.
>
> Someone should try this on tux2[1] / ext3[2] / reiserfs[3]

I tried it on my home box, running Linux 2.2.17 + reiserfs 3.5.27.

(AMD K6-2/300, VIA VT82C586B, Western Digital Caviar AC420400D in
UDMA/33 mode)

I created an fstest directory, ran the above fsstone command and pressed
reset after a few seconds and rebooted.

The journal replay took 12 seconds for 127 transactions, after that, I
had 688 files in that directory.

I also had reiserfsck 3.5.27 check that file system which found no
errors.

That's - however - just one sample. It's not on SCSI, it does not check
this or that other.

I also benchmarked time fsstone -cr 2000 5000:
                                               real user syst
On async ext2fs (default), fsstone took 25 s. 37.9 4.4 28.9
On sync ext2fs (chattr), fsstone took 29 s. 52.7 4.8 31.5
On reiserfs (tea hash), fsstone took 40 s. 137.7 4.4 17.4

-- 
Matthias Andree