Re: Operation Questions

From: Miguel A.L. Paraz <[email protected]>
Date: Mon, 10 Feb 1997 16:09:00 -0800 (GMT+8)

Hello,

Tim Carlier wrote:
> 1) Say I have 128M of RAM, I believe the recommended setup is to
> configure Squid to use about 1/3 of this for it's primary memory pool -
> does it's memory usage grow and grow beyond physical RAM and then into
> swap (if needed) or does it just limit itself somehow?

It uses swap; it does not limit its usage.

> What would be the ideal memory config parameters if using 128MB?

Try the 1/3 value; beyond that, it's to each his own?
 
> 2) If I have the choice of configuring several disk partitions as either
> a single multi-device as in Linux's /dev/md or leave them single, which
> would give me better performance (if any difference)? If I configured
> several partitions using RAID-0 under MD, would this be better than
> telling squid to use several disk partitions?

Are your separate partitions on separate physical hard disks?

If so, yes I think it would benefit (though I haven't had the chance to
try it), especially if the different partitions are of different sizes.
Currently all the cache_dir partitions need to be of the same size.
 
> 3) As the cache becomes full, does Squid automatically begin to purge
> the oldest data kept, or what stops it from overflowing?

Check squid.conf's cache_swap_low and cache_swap_high parameters,
as well as reference_age.
 
> 4) If I ever wanted to - what's the best way to purge the whole cache or
> clean it up; is there a special Squid function or is it just done by
> nuking the cache dir?

There's no delete function; nuking the cache dir via mkfs or mke2fs
would be faster!

Regards,

-- 
miguel a.l. paraz  <map@iphil.net>              |  pgp key id: 0x43F0D011
iphil communications, makati city, philippines  |  <http://www.iphil.net> 
Received on Mon Feb 10 1997 - 00:27:27 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:34:24 MST