[Gllug] Re: Memory usage

[Nix]

On Thu, 13 Oct 2005, Peter Grandi announced authoritatively:
>>>> On Thu, 13 Oct 2005 18:36:53 +0100,
>>>> pg_gllug at gllug.for.sabi.co.UK (Peter Grandi) said:
> 
> [ ... ]
> 
> pg_gllug> Whatever, except for VIM above we see lots and lots of
> pg_gllug> attempted or successful inode accesses. If those
> pg_gllug> inodes are scattered around the disc, very bad news.
> 
> BTW, to forestall at least one case of inevitable and pointless
> nixpicking, yes, many (but not most) of the many hundreds inode
> accesses above, especially the 'stat64's (which is used to do
> 'access(3)') are repetitive so even from cold start they will be
> cached after the first.
> 
> Anyhow, the unique accesses to possibly far flung inodes that
> remain are still an obscene amount.

inodes are very aggressively cached (far more aggressively than file
data by default). I find that if I stat() *every file on my local disks*
it stays in memory for half an hour or so on a system with 512Mb RAM.

(Obviously said system wasn't under severe memory pressure: if it was,
the cache would be discarded, of course.)

-- 
`Next: FEMA neglects to take into account the possibility of
fire in Old Balsawood Town (currently in its fifth year of drought
and home of the General Grant Home for Compulsive Arsonists).'
            --- James Nicoll
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug