Re: Performence problem in lazarus

No work has been done on lazarus in several years (& I personally wouldn't
use the proposed solution, but c'est la vie ;-)).  I believe that brian
carrier's Autopsy (from http://www.sleuthkit.org/autopsy/desc.php)
implements something that addresses this serious issue.

-- d

> [email protected] sez:
> 
> 	Does the new version take care the following problem and
> 	the hash solution?
> 
> Bye,
>   Udi
> 
> 
> On Wed, 31 May 2006, Udi Mottelo wrote:
> 
> >
> >THE PROBLEM:
> >
> >	I try to recover some files from big filesystem (18Gb from 30Gb)
> >	and I find that after 10 hours the load everage is intolerable
> >	and I must kill lazarus.  The computer is strong HP-dc7600 with
> >	1Gb memory and does nothing but lazarus.
> >
> >THE REASON:
> >
> >	Lasarus makes one flat directory (the bolcks dir) and too much
> >	inodes in this directory make the over head.  Simple command
> >	`ls blocks`   takes more then 15 minuts !!!!!
> >
> >THE SOLUTION:
> >
> >	I suggest to adopt the terminfo way or some mail spool way ie
> >	split the blocks dir to subdirectories by prefix of the filename.
> >	In following example I use the first four charachters and reduce
> >	the number of inode/dir in 10^3:
> >
> >	the file 0000001 will be saved in subdirectory 0000
> >	the file 0000002 will be saved in subdirectory 0000
> >	...
> >	the file 0001295 will be saved in subdirectory 0001
> >	...
> >	the file 0002004 will be saved in subdirectory 0002
> >	...
> >	the file 9998749 will be saved in subdirectory 9998
> >
> >
> >	Unfortunately I'm not good programer and I don't know PERL, so
> >	I'll appreciate if somebody will help me to make the changes
> >	because right now I'm stuck.  I hope that it will help more users
> >	of TCT because in this days the	filesystems became more bigger and
> >	this patch can be real improvement.
> >
> >
> >Thanks in advanced,
> > Udi
> >
> >+++++++++++++++++++++++++++++++++++++++++++
> >This Mail Was Scanned By Mail-seCure System
> >at the Tel-Aviv University CC.
> >