Date: Thu, 10 Sep 1998 09:54:52 +0930 From: Greg Lehey <grog@lemis.com> To: Mike Tancsa <mike@sentex.net> Cc: questions@FreeBSD.ORG Subject: Re: vinum (was Re: Software RAID) Message-ID: <19980910095452.K583@freebie.lemis.com> In-Reply-To: <35f71995.77963685@mail.sentex.net>; from Mike Tancsa on Thu, Sep 10, 1998 at 12:16:29AM %2B0000 References: <19980909182749.N583@freebie.lemis.com> <35f71995.77963685@mail.sentex.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday, 10 September 1998 at 0:16:29 +0000, Mike Tancsa wrote: > On Wed, 9 Sep 1998 18:27:49 +0930, in sentex.lists.freebsd.questions > you wrote: > >> On Tuesday, 8 September 1998 at 1:37:49 +0000, Mike Tancsa wrote: >>> On Mon, 07 Sep 1998 23:22:05 +0100, in sentex.lists.freebsd.questions >>> you wrote: >>> >>>> Hello, >>>> >>>> does or will FreeBSD support software RAID? >>> >>> Partially... Have a look at >>> >>> man -k ccd >> >> Vinum will replace ccd. > > Would you say its ready for prime time in for example a new server > environment ? Good question. The current state is still "alpha", and there's a lot of stuff that still needs to be written, and other parts haven't been tested to my satisfaction: for example, I'm not convinced that recovering defective volumes on line works correctly in all cases. It compares favourably to ccd, though: in ccd, if a you have a drive failure, you have to modify your ccd config and reboot before you can access a mirrored pair again. Vinum handles this transparently. As far as reliability goes, a number of people have been using it heavily for a month now, and I haven't had any problem reports in that time (I *have* had reports of "works fine, no problems"). > Also, on my news server I have a few drives that are of different > make and model. Last time I looked at ccd it was not reccomended to > try and stripe drives of different sizes and performance specs. Hmm. I can't think why not. > Can I use vnum to make 1 virtual news spool drive? Yes. > Also, how well does vinum cope with disk errors. I have one fuji > thats somewhat prone to them... It's an area I want to improve on. It handles write errors fine, but there's a possibility it will return a transient error on a read (i.e. the read will fail, but a retry from the application will succeed). The correct action, of course, is that the read will not fail if a retry would succeed. ccd is also prone to this problem, which is related to the fact that both ccd and Vinum read from only one disk, and they can't do retries at the point where they discover the error (in the bottom half of the driver). It's possible to fix this problem, but due to the kernel structure it's relatively complex. I'd expect it to be in by the end of the year, though. Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980910095452.K583>