Date: Sat, 17 Jan 2004 19:45:18 -0500 From: Jeremy Faulkner <gldisater@gldis.ca> To: Colin Percival <colin.percival@wadham.ox.ac.uk> Cc: freebsd-chat@freebsd.org Subject: Re: Good BSD/Linux Article (somewhat off-topic) Message-ID: <4009D71E.3020209@gldis.ca> In-Reply-To: <6.0.1.1.1.20040118000417.02bbee70@imap.sfu.ca> References: <6.0.1.1.1.20040116175159.03f4dd48@imap.sfu.ca> <Pine.NEB.3.96L.1040117185613.22159B-100000@fledge.watson.org> <6.0.1.1.1.20040118000417.02bbee70@imap.sfu.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
Colin Percival wrote: > At 23:59 17/01/2004, Robert Watson wrote: > >> I suspect that the /. effect has gotten easier to carry >> over time in part because a lot of the clients are higher bandwidth than >> they were before -- if you have moderate size files being tranfered, lots >> of long-lived slow connections take up a lot more memory than short-lived >> ones. > > > Actually, this raises an interesting point -- if > 1. There is a significant amount of network traffic, > 2. There is memory pressure, and > 3. There are several runnable processes, > it might be a good idea to give scheduling priority to the oldest > process, in the hope that it will complete and free its memory. > > Colin Percival dnetc and seti would be the oldest process on some machines. So making this a mandatory setting would be counter productive. -- Jeremy Faulkner http://www.gldis.ca
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4009D71E.3020209>