Date: Fri, 29 Jun 2007 10:14:07 +0300 From: Nikos Vassiliadis <nvass@teledomenet.gr> To: freebsd-questions@freebsd.org Cc: Pieter de Goeje <pieter@degoeje.nl>, Shawn O'Connor <soconnor@falconknight.com> Subject: Re: mdconfig using malloc() Message-ID: <200706291014.08311.nvass@teledomenet.gr> In-Reply-To: <DCA423A8-A657-4EB0-A252-48B79FFAF903@falconknight.com> References: <4D15FF0C-0531-4455-9631-F60D939F9EB2@falconknight.com> <200706290141.07063.pieter@degoeje.nl> <DCA423A8-A657-4EB0-A252-48B79FFAF903@falconknight.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Friday 29 June 2007 03:01, Shawn O'Connor wrote: > >> I saw some old post advocated using -t swap instead, saying that swap > >> probably wouldn't be used until real memory was consumed. But, on my > >> 6.2 machine that immediately raises the swap usage to 2GB and does > >> use disk instead of RAM when I test using it. > > > > That's because you used -o reserve. Leave it out and it'll be fine. > > When using with -o reserve and swap it returns immediately, just like > with -t malloc without the -o reserve (except without the nasty > repercussions of running out of RAM). My problem isn't that it uses > the swap (nothing is using it anyway), but that it is using disk at > all, which is way to slow for what I'm trying to do. Why do you even have swap? Is it for kernel dumps? Why don't you simply eliminate the swap device? But, as Pieter also suggested, I believe tmpfs would be much more efficient in every regard. Perhaps not stability-wise, though the code is new in the FreeBSD src tree, it has been in the NetBSD src tree since right after SoC2005. Nikos
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200706291014.08311.nvass>