Date: Mon, 26 Jul 2010 18:53:48 +0100 From: RW <rwmaillists@googlemail.com> To: freebsd-hackers@freebsd.org Subject: Re: pageout question Message-ID: <20100726185348.63ebf916@gumby.homeunix.com> In-Reply-To: <4C4CA1DC.2050902@freebsd.org> References: <4C4B4BAB.3000005@freebsd.org> <20100725003144.3cfead39@gumby.homeunix.com> <4C4C0CD9.6000002@freebsd.org> <20100725144141.6f1f33cc@gumby.homeunix.com> <4C4C47FD.6080802@freebsd.org> <20100725212849.1e07f40c@gumby.homeunix.com> <4C4CA1DC.2050902@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 25 Jul 2010 23:43:08 +0300 Andriy Gapon <avg@freebsd.org> wrote: > on 25/07/2010 23:28 RW said the following: > > I didn't say it say it was guaranteed. I just think the scenario > > where a first pass ends up between the watermarks is rare. And when > > it happens I don't see a compelling reason to do extra paging to > > reach an arbitrary target. > > Well, it seems neither I nor you have data to show whether it's rare > or not (and it would greatly depend on workload too). > As to "arbitrary target" - well, that's the whole point of > hysteresis-like behavior. We start paging also at an "arbitrary" > point. If after the first pass with light-paging the high watermark isn't reached then the choices are 1) loop and immediately do a heavy-paging pass. 2) wait and let the daemon get woken-up for another light-paging pass - only go to heavy-paging when this strategy isn't keeping up with demand. To me (2) is doing the right thing. It's trying to satisfy demand from existing clean pages, and only paging heavily as a last resort.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100726185348.63ebf916>