Date: Wed, 06 Mar 2002 15:19:42 -0600 From: Craig Boston <craig@meoqu.gank.org> To: stable@freebsd.org Cc: sos@freebsd.dk, grog@freebsd.org Subject: Vinum on ATA-RAID devices (was Re: Request for testers of new ATA driver patches) Message-ID: <3C8687EE.9060304@meoqu.gank.org> References: <200203060821.g268LR045942@freebsd.dk> <3C864253.5000700@meoqu.gank.org>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Craig Boston wrote:
> (blah blah blah)
Okay, this is the last post to the list on this topic. I promise :)
It appears my previous patch is not quite enough. It gets vinum working
all right, but every time I reboot it "forgets" the config and has to be
re-created. I tracked the problem down to the fact that ar devices
aren't in the list provided by getdevs(), and that's how vinum figures
out where to load its configuration from.
So, the attached patch should fix both problems. I've been running it
for the past few hours (along with the backported -current ATA driver),
and so far everything is working flawlessly and has survived a couple
buildworlds while my mirror is rebuilding. vinum ld shows:
D alpha State: up Device /dev/ar0s1e Avail:
1/18600 MB (0%)
D beta State: up Device /dev/ar1s1e Avail:
1/18600 MB (0%)
Yay! :) The more I think about it, the more I see the usefulness of
vinum on ar devices. For example, if your RAID controller lets you
create new arrays but not expand existing ones, you could add it to a
vinum span set to simulate the effect (I've had to do this with Compaq
SCSI RAIDs).
Sren: It's your driver -- is there any particular reason that ar should
be left off the devstat list? My patch adds a devstat entry for it, but
never removes it. As far as I can tell disk_destroy never happens for
ata-raid devices, so I think/hope this is the correct behaviour. It
also doesn't implement devstat transactions, but I'm not sure if that
even makes sense due to ar being more of an abstraction. I don't know
yet if there will be any ill effects from doing this as I'm not
incredibly familiar with the ATA subsystem.
The patch is for -stable systems with Soren's -current ATA driver
backport already applied. I also have a version for stock RELENG_4
systems with the old driver, but am not posting it because it can cause
problems due to vinum being able to access the subdisks directly and it
reading them first. I doubt the patch will apply cleanly to -current,
but with a few minor changes should be able to work...
Thanks -- I'll stop bugging you guys now :)
Craig
[-- Attachment #2 --]
v< ar-vinum-2002-0306.patch Smo0 ~ũ &
H«`NѴQhS2S9 ] }=a? w ZPFF|` V1N4F L
Ӯ0)CS @/J MӲ0y1ִN 9"K /(, թ0%AԠGgQQc=K?fBf_Q^)S8oKx^2( M~BB< 5M!:LNbq'cT 2p80ˠmLnqbKgzX#쿢9ƶ sxr,%&
ҒU褂/I=gM:F[wPƈT
T#uJT+yr/tjZ3|Y7_x9g W=vA殷oyL\O2}
&LL$P1*aER+zqZA&v2"ey=?ֱ⺸;lqR(+enYu/z=<Q.Q-$> X֠RxsBhdd)M8P4&T {>>Sϳ+/̟7s)ֲ/f Ya]z bT0
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C8687EE.9060304>