Date: Wed, 11 Jan 2006 06:58:34 -0500 (EST) From: "Brian Szymanski" <ski@mediamatters.org> To: "Stijn Hoop" <stijn@win.tue.nl> Cc: freebsd-stable@freebsd.org, Lukas Ertl <le@FreeBSD.org> Subject: Re: gvinum/vinum on 6.0 Message-ID: <3657.68.49.189.193.1136980714.squirrel@68.49.189.193> In-Reply-To: <20060111064349.GJ63938@pcwin002.win.tue.nl> References: <2178.68.49.189.193.1136950709.squirrel@68.49.189.193> <20060111064349.GJ63938@pcwin002.win.tue.nl>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] Stijn, thanks for your help, I'm getting closer... >> I took 6.0 for a test drive today and was disappointed to find that >> vinum/gvinum are still in disarray. For example there is a man page for >> vinum, but only a gvinum binary. gvinum help still lists lots of old >> vinum >> commands that are not implemented in gvinum. Lots of basic things I try >> from the gvinum prompt just tell me "not yet supported". > > Hmm. There is a manpage in 6-STABLE. And there are a few things that > don't work but I wouldn't call it "lots". Ah, a manpage! Progress... >> But most importantly, gvinum configuration (at least for a raid-5 plex) >> still doesn't persist across a reboot :( > > That's a bug; I think it might be related to compiling gvinum in the > kernel > as opposed to loading it from /boot/loader.conf. I also think there is a > fix already commited to 6-STABLE. Hmm, I upgraded to 6-STABLE and I'm still having the problem. Here's basically how it happens: gvinum create /etc/vinum.cnf newfs /dev/gvinum/VOLUME mount /dev/gvinum/VOLUME /mnt #screw with /mnt, everything works and is happy, yay! reboot At this point I call "gvinum l" (which loads geom_vinum.ko) by hand (after the reboot). My configuration mostly seems to persist - except or the "drives" section... One of two things happens here: Either a) The volumes/plexes appear down, the subdisks are stale. Additionally, /dev/gvinum is not there - suffice it to say I can't mount anything. b) Everything has status up (except the nonexistent drives). In this case as well, /dev/gvinum is not there - and I still can't mount anything. If I try to fix the configuration by gvinum rm'ing the volumes and plexes that are there, then reloading my vinum.cnf (gvinum create), either: a) everything seems to work b) the system panics Unfortunately there is no correlation between "gvinum l" output and whether the system panics or is happy when I run gvinum create again. Would I have better luck compiling gvinum into the kernel instead of loading the module? What do other folks who have successful config's do? Thanks in advance, Brian Brian Szymanski Software and Systems Developer Media Matters for America ski@mediamatters.org [-- Attachment #2 --] drive apple device /dev/ad1d drive banana device /dev/ad2d drive orange device /dev/ad3d volume rv plex org raid5 288k sd length 1018m drive apple sd length 1018m drive banana sd length 1018m drive orange [-- Attachment #3 --] 3 drives: D orange State: up /dev/ad3d A: 6/1023 MB (0%) D banana State: up /dev/ad2d A: 6/1023 MB (0%) D apple State: up /dev/ad1d A: 6/1023 MB (0%) 1 volume: V rv State: up Plexes: 1 Size: 2035 MB 1 plex: P rv.p0 R5 State: up Subdisks: 3 Size: 2035 MB 3 subdisks: S rv.p0.s2 State: up D: orange Size: 1017 MB S rv.p0.s1 State: up D: banana Size: 1017 MB S rv.p0.s0 State: up D: apple Size: 1017 MB [-- Attachment #4 --] 0 drives: 1 volume: V rv State: down Plexes: 0 Size: 0 B 1 plex: P rv.p0 R5 State: down Subdisks: 0 Size: 0 B 3 subdisks: S rv.p0.s0 State: stale D: apple Size: 1017 MB S rv.p0.s1 State: stale D: banana Size: 1017 MB S rv.p0.s2 State: stale D: orange Size: 1017 MB
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3657.68.49.189.193.1136980714.squirrel>