Michael Baird wrote:
>
> On my experimental server, I've recently changed the maximum_object_size
> from the default (4 MB) to 32MB to try to bring up the bytes_hit ratio,
> I'm using 2.4 dev4 w/DISKD as well as LRU replacement, what I've noticed
> with this being the only change made, is now a mostly negative byte hit
> ratio, server_kbytes > client_kbytes? Is this something I should be
> concerned with, or do I use a different replacement policy, or some
> other settings I need to change, frankly I expected my byte_hit ratio to
> go up, not the opposite.
I've also detected the same thing when testing 2.4DEVEL4,
but I didn't relate that to the maximum object size, only
to the change from DEVEL2 -> DEVEL4 (both using diskd). I had
BHRs on the hour well bellow 50%. I tried out LRU and LFUDA.
It's a mistery to me...
--jp
>
> Regards
> MIKE
>
> --
> To unsubscribe, see http://www.squid-cache.org/mailing-lists.html
-- FCCN - Funda��o para a Computa��o Cient�fica Nacional - Tel: +351-218440100 -- To unsubscribe, see http://www.squid-cache.org/mailing-lists.htmlReceived on Fri Sep 29 2000 - 12:23:14 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:55:31 MST