Date: Fri, 9 Sep 2011 14:40:03 GMT From: Jonathan <jonathan@kc8onw.net> To: freebsd-bugs@FreeBSD.org Subject: Re: kern/157293: [mfi] mfiutil/mfi does not allow adding a previously configured drive to be added to a new array Message-ID: <201109091440.p89Ee3EO040964@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
The following reply was made to PR kern/157293; it has been noted by GNATS. From: Jonathan <jonathan@kc8onw.net> To: Garrett Cooper <yanegomi@gmail.com> Cc: bug-followup@FreeBSD.org Subject: Re: kern/157293: [mfi] mfiutil/mfi does not allow adding a previously configured drive to be added to a new array Date: Fri, 09 Sep 2011 10:20:26 -0400 On 9/9/2011 4:18 AM, Garrett Cooper wrote: > I wish I had seen this issue earlier so I could have sent a suggestion > (in all fairness this should have been a post to an email list, like > freebsd-scsi, freebsd-stable, or even freebsd-questions. Noted. > If you have a degraded stripe-only RAID, then you're basically SOL. That > aside.. if you didn't use a straight RAID-0, you could have readded the > new drive as a hot spare like so: > mfiutil add <drive-number> <volume> > and the controller would have added the drive to your RAID. The drive in question was the only member of it's RAID stripe because I could not get the controller to do pass through on it when I first set it up. For some reason the drive was not properly recognized after a power failure and in the process of trying to walk the data center through bringing in back over the phone it was somehow removed from the controllers stored configuration. The drive itself still had the LSI metadata on it somewhere that made the controller think it was still in use elsewhere so it refused to let me create a new RAID or JBOD device with it. Would the "add" command have still worked in that case, it seems unlikely? The original drive turned out to be fine. I got it from the datacenter, put it in my home server, and imported the ZFS pool without a problem and with all of it's data intact and it's still there now. It seems like there needs to be a way to either force a drive into an mfi RAID or to somehow wipe the controller metadata without having to dd /dev/zero to the drive.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201109091440.p89Ee3EO040964>