Date: Wed, 23 May 2007 12:44:50 +0930 From: "Daniel O'Connor" <doconnor@gsoft.com.au> To: "Bruce M. Simpson" <bms@incunabulum.net> Cc: freebsd-stable@freebsd.org Subject: Re: Gigabyte GA-VM900M caveats Message-ID: <200705231245.06577.doconnor@gsoft.com.au> In-Reply-To: <4653AB59.2020702@incunabulum.net> References: <46535A83.5050207@incunabulum.net> <200705231108.44939.doconnor@gsoft.com.au> <4653AB59.2020702@incunabulum.net>
next in thread | previous in thread | raw e-mail | index | archive | help
--nextPart2274876.Y8xyHxjRZx Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On Wednesday 23 May 2007 12:17, Bruce M. Simpson wrote: > Thanks for the affirmation that I'm not alone in this. The whole > point of RAID being supported by platform firmware is to facilitate > booting from it even if disks fail. Yes, I was pretty pissed off when I found out about it. I was even more=20 pissed off at Epox's weak response :( > > When I asked my motherboard vendor about this they suggested I > > switch to SATA mode (to boot off one disk) or to put another disk > > in and rebuild... Not exactly good for unattended use. > > Summary: Avoid VIA software RAID like the plague. Yep. The JMicron RAID in my Core 2 Duo system seems OK.=20 That said all of these software RAID implementations seem to suffer from=20 a serious limitation - if the disk fails temporarily and the array=20 becomes degraded and you then subsequently reboot it will see 2 arrays=20 and can boot off the stale one. I would expect a more sensible implementation where it would scan the=20 disks and if it sees 2 disks with the same array ID but differing ages=20 it should ignore the older disk. This assumes that the RAID code increments the generation number (or=20 whatever) when it notes the array is degraded.. I would hope so but=20 maybe I'm wrong. I have been bitted by this several times especially in SATA as it seems=20 more likely to get temporary failures (especially with dodgy PHYs like=20 Marvell) > Yes, utter crap, and a waste of valuable time. Yeah, I stumped up for 3ware in the end. Not that they are MUCH better, I have several pending issues with them, =20 but so far as I can see they are the only manufacturer of 2 port=20 hardware RAID cards. > [Normally I have been running 7-CURRENT on that machine, with the > JMicron card, so up until now this hasn't been an issue, but this is > what kicked off the whole shooting match in the first place, and I > need to be able to multi-boot Windows "Longhorn" Server and Gentoo > Linux for the work I'm going to be doing.] I have Windows XP and FreeBSD 6.2 on my Core 2 Duo/JMicron system. sos@=20 is looking at a problem in current (it doesn't see my disks). I tried Ubuntu on this system but its dmraid doesn't grok the JMicron=20 metadata properly (not to mention that dmraid does almost no error=20 handling and so I am loathe to use it) > I wonder if people have had better experiences with JMicron. I am > encouraged by the work Scott Long has begun, although that is going > to take time to bear fruit. I can't burn too much time on this > though, I needed a working server today. I only have 1 JMicron system and it works OK so far (modulo my issue=20 in -current). The last good, cheap RAID card I used was a Promise FT100/TX2, alas it=20 is PATA only and they don't sell them any more. =2D-=20 Daniel O'Connor software and network engineer for Genesis Software - http://www.gsoft.com.au "The nice thing about standards is that there are so many of them to choose from." -- Andrew Tanenbaum GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C --nextPart2274876.Y8xyHxjRZx Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (FreeBSD) iD8DBQBGU7G65ZPcIHs/zowRAh23AJ9/Bv/0oZSwUlpXozK0ZqmsMvIwIwCeOdr0 9nXFMGlMx467o0+RqcmedUA= =gFR9 -----END PGP SIGNATURE----- --nextPart2274876.Y8xyHxjRZx--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200705231245.06577.doconnor>