Date: Mon, 26 Aug 1996 13:49:34 -0500 (CDT) From: Joe Greco <jgreco@brasil.moneng.mei.com> To: michaelv@MindBender.serv.net (Michael L. VanLoon) Cc: jgreco@brasil.moneng.mei.com, michael@memra.com, freebsd-isp@freebsd.org, craigs@os.com Subject: Re: Anyone using ccd (FreeBSD disk striper) for news Message-ID: <199608261849.NAA00979@brasil.moneng.mei.com> In-Reply-To: <199608261830.LAA16588@MindBender.serv.net> from "Michael L. VanLoon" at Aug 26, 96 11:30:49 am
next in thread | previous in thread | raw e-mail | index | archive | help
> >But async mode will not buy you much in this case...! Async is simply > >delaying the inevitable, which is already delayed in FreeBSD until the next > >30-second sync. I do not want to write metadata at _all_ if all it is is > >an ATIME update... if I have to write it, now OR later, I am losing > >bandwidth and transactions per second that I would rather spend reading > >other data. > > Not completely. What if the data that is being postponed changes two > or three times while it's in the cache. When it finally gets written, > you just wrote once what would have been written two or three more > times. First thing, on a news server, it ain't terribly likely to happen. Second thing, that already happens, because FreeBSD only flushes once every thirty seconds. inode updates are already async. What I want is a solution which postpones the writes until hell freezes over. Or, better yet, doesn't do them at all. ... JG
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608261849.NAA00979>