Date: Thu, 08 Apr 1999 15:13:29 +0800 From: adrian@freebsd.org To: freebsd-hackers@freebsd.org Cc: Markus Stumpf <maex-freebsd-hackers@Space.Net> Subject: Re: squid and FreeBSD (was: Re: Nice FreeBSD mention) Message-ID: <19990408071330.5268.qmail@ewok.creative.net.au> In-Reply-To: Your message of "Thu, 08 Apr 1999 00:18:50 MST." <199904080718.AAA09117@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Matthew Dillon writes: > >:"top" says >: >: PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND >: 8542 root 2 -15 204M 177M select 60:27 0.00% 0.00% squid >: >:>From that point I would imagine that 256 MB RAM is enough to fit in >:squid and - with about 50 megs left - some other processes. > > I'm not familiar with squid. Why is it eating 204M of VM? If that > is programmable, simply reduce it to 150M. > > FreeBSD can't give over all of its physical memory to a process. It > needs memory for kernel buffers and file caching. It *especially* > needs memory for kernel buffers in a network-heavy setup. > > Turning off swap in this case will cause the machine to lockup when > it runs out of memory -- probably not what you want. > Drop your cache_mem to something really small if possible. Let the OS do the file buffering and you'll find it should swap less. If squid starts complaining the cache_mem is way way too small, its time for FreeBSD is probably paging out unused pages in squid, and since most of squids footprint will be the cache object database, on a busy squid it will result in lots of pageins, causing degraded performance. Adrian To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990408071330.5268.qmail>