RE: cache layout questions (& disaster recovery!)

>> According to the squid FAQ it is much better to spread cache directories
>> over several physical disks and even the controllers (!).

Sure, that makes sense. If I had the choice of buying the hardware for
this machine I'd get a few 1GB disks for it and a couple of controllers,
but as it stands I'm stuck with just a pair of large disks with which to
work, so the only question about one big cache or lots of little ones is
whether having several small caches allows squid to find cached files
quicker because the individual cache.log files will be smaller (or does
it still use one global cache log file??)

>> - raise squid process priority during startup by starting it with command:
>> "nice -j -20 /usr/local/squid/bin/squid -Y"

fair enough. Squid's not using much by way of CPU resources on the
current system we have though, disk/memory/network-throughput will be
the main problems I expect. If I can get two token ring cards working
together in the system (wish we were on ethernet here!!) then that
should sort out any networking bottlenecks, 256MB of memory in the 'new'
system (as opposed to the 64MB in the current one) should solve that
problem, so the disks are the only hassle at the moment.

Having said that, RedHat 5.2 is having a total spasm trying to work with
the Adaptec 2940UW controller I have at the moment - I don't know if the
driver is broken or whether it's a cable/hardware problem but I'm
getting timeouts on the SCSI bus all over the place, so until I solve
that one I'm stuck anyway :*)

cheers,

Jules

ps. let us all know when that script's ready - sounds most useful!!
>
Received on Fri Jun 11 1999 - 04:42:00 MDT