Date: Mon, 31 Aug 2015 07:00:18 -0600 (MDT) From: Warren Block <wblock@wonkity.com> To: hackers@freebsd.org Subject: Re: Puzzle involving serial console and ...??!? Message-ID: <alpine.BSF.2.20.1508310650420.42983@wonkity.com> In-Reply-To: <20150831030639.GH1153@albert.catwhisker.org> References: <20150831030639.GH1153@albert.catwhisker.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 30 Aug 2015, David Wolfskill wrote: > I may be going mad -- that's likely the simplest explanation, and > Occam's Razor would thus indicate that this hypothesis should be > accorded a degree of plausibility, at least. > > As described in <http://www.catwhisker.org/~david/FreeBSD/upgrade.html>, > I have my own quirky ways of upgrading FreeBSD systems, the basics > of which remain fundamentally unchanged for the last dozen years > (well, a few more than that, but...). In turn, much of that approach > was borrowed from one of my pre-FreeBSD experiences in an IBM > mainframe environment (as a "systems programmer"). > > What's causing me to question my tenuous grasp of reality is this: > Referencing the above-cited Web page, my build machine ("freebeast") > normally runs headless, and I use a cable to connect its serial > port to the serial port of one of my other machines ("albert"). > > By doing so, I have a working serial console (once the FreeBSD > loader gains control from the BIOS, at least)... half the time. > > I have determined that whether or not I have a functioning serial > console for freebeast depends on whether I boot albert from slice > 1 or slice 2: if from slice 1, it doesn't work; if from slice 2, > it does. > > Even when the contents of the 2 slices are functionally equivalent > (that is, the file systems of the 2 slices are as "the same" as a > "dump 0Lf | restore" pipeline can make them, except for a deliberate > difference in the contents of /etc/fstab (as the root and /usr file > systems are on different slices in the two cases)). Unless there is a newfs in there also, there could be additional files on the target filesystems. Maybe an mtree comparison or rsync -n would reveal differences. "nodump" flags and dump's -h flag is another possibility, although less likely.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.20.1508310650420.42983>