time spent writing history..

[Fabien Tassin]

According to Alex Kiernan:
> 
> Fabien Tassin <fta at sofaraway.org> writes:
> 
> > I wonder if someone has the same problem or if it is only me..
> > 
> 
> We're seeing a similar problem on a development box - we're testing a
> number of storage solutions & the current hypothesis is that we
> started seeing performance problems when we shifted the history db to
> a filer from local disk.

The numbers I've given are for a local history db. path_db is on a (hard) RAID1
array built with Ultra3 10krpm SCSI disks on a Ultra2 (AIC-7880 + MegaRAID)
controler. The kernel is currently a Linux v2.4.0 (soon to be a 2.4.1 to fix
a nasty driver bug).
As expire is damn fast, I assume it is not an hardware problem. 

> The patch looks like its just what I need to help me answer that!

If you try it, please, show me your reports.

-- 
Fabien Tassin -+- fta at sofaraway.org