Date: Sat, 25 Sep 2004 19:07:16 -0400 From: Chris Laverdure <dashevil@sympatico.ca> To: Peter Jeremy <PeterJeremy@optushome.com.au> Cc: freebsd-current@freebsd.org Subject: Re: panic: swap_pager_isswapped: failed to locate all swap meta blocks Message-ID: <1096153635.937.1.camel@elemental.DashEvil> In-Reply-To: <20040925200747.GE83620@cirb503493.alcatel.com.au> References: <7m7jqjhojv.wl@black.imgsrc.co.jp> <20040924122508.GG9550@darkness.comp.waw.pl> <20040924143224.GG47816@dan.emsphone.com> <20040925095549.GH9550@darkness.comp.waw.pl> <20040925200747.GE83620@cirb503493.alcatel.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 2004-09-25 at 16:07, Peter Jeremy wrote: > On Sat, 2004-Sep-25 11:55:49 +0200, Pawel Jakub Dawidek wrote: > >On Fri, Sep 24, 2004 at 09:32:30AM -0500, Dan Nelson wrote: > >+> Considering that the system is going down anyway, why bother? > > > >Because if you have swap on e.g. gmirror device, gmirror will rebuild > >components on every boot, because components are marked dirty - > >they were open for writing on shutdown, so it looks like a power failure. > > As an alternative approach, rather than marking swap clean on a > shutdown, why not have a flag in the object's metadata that says "this > object doesn't need synchronising on a reboot". If the flag is set, > then gmirror just sets both sides as synchronised and active on boot. > This is the approach taken by HP Tru64 LSM. (Though one improvement > you could make over LSM would be to document the flag). Couldn't you just kill all the process' and then (since everything else should be wired memory anyway) just disable swap without even bothering to swap it back into RAM? I mean, you are shutting down anyway. Who cares if the programs are in memory or not, they are supposed to be shut down.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1096153635.937.1.camel>