Date: Wed, 25 Aug 2010 17:53:25 +0300 From: Andriy Gapon <avg@icyb.net.ua> To: freebsd-fs@freebsd.org Subject: Re: zfs arc - just take it all and be good to me Message-ID: <4C752E65.3070701@icyb.net.ua> In-Reply-To: <20100824211136.GA18104@tolstoy.tols.org> References: <20100810214418.GA28288@tolstoy.tols.org> <20100811014919.GA52992@icarus.home.lan> <20100811192537.GA44635@tolstoy.tols.org> <AANLkTin-YvEzoN-ThwwDAqn2mWFMD4-7BnP8N95EqTk0@mail.gmail.com> <20100811214302.GB44635@tolstoy.tols.org> <20100812205625.GA79515@server.vk2pj.dyndns.org> <20100824150035.GB99477@roberto-al.eurocontrol.fr> <20100824200527.GC11990@server.vk2pj.dyndns.org> <20100824211136.GA18104@tolstoy.tols.org>
next in thread | previous in thread | raw e-mail | index | archive | help
on 25/08/2010 00:11 Marco van Tol said the following: > Would the inactive pages of the long running processes be swapped out in > favour of the arc? That would be great I think as long as that is made > clear at an obvious place so people don't get worried about swap usage > on a system that "shouldn't need to swap out anything". > > /usr/src/UPDATING springs to mind, together with quick hints on how to > see that the swapped out pages are harmless. :) I don't think there would be any significant impact on swapping. Remember that inactive pages also include clean pages that can be moved to cache and vnode pages that can be synced to disk. pagedaemon is smart enough to not trigger swapping too early. Makes sense? :) -- Andriy Gapon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C752E65.3070701>