Date: Wed, 28 May 2014 15:15:05 +0000 From: Paul J Murphy <paul@inetstat.net> To: freebsd-geom@freebsd.org Subject: Re: conf/190152: [patch] [rc] gmirror savecore support Message-ID: <20140528151505.GA42484@srv00.inetstat.net> In-Reply-To: <20140528111156.GA42016@srv00.inetstat.net> References: <5c0c75150aa4e34ba25db65258052128@lhaven.homeip.net> <20140528111156.GA42016@srv00.inetstat.net>
next in thread | previous in thread | raw e-mail | index | archive | help
http://www.freebsd.org/cgi/query-pr.cgi?pr=conf/190152 Hi geom-people, a friendly heads-up on the above PR. I'm seeking validation of my belief that there's no changes necessary to /etc/rc.d/dumpon, only /etc/rc.d/savecore. The kernel will always deterministically pick a single physical device within the mirror, and we just need to use "prefer" to force the same determinism when savecore runs, right? (With the obvious caveat that failed and replaced devices, or other changes to the devices and their priorities that make up the mirror, will change the results of the determinism.) Any feedback on my patch in conf/190152 is welcome. Its aim is to at least cover the common and simple cases with gmirrored system disks. See also http://www.freebsd.org/cgi/query-pr.cgi?pr=docs/178818 which initially was attempting the fix via dumpon alone. Thanks, Paul.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140528151505.GA42484>