Re: NetBSD RAIDFrame issue (kind of off-topic).

[buhrow%lothlorien.nfbcal.org@localhost (Brian Buhrow)]

        hello dave.  I agree with  Greg with the following additional
questions/observations.

1.  You don't say, but I assume you re disklabled/partitioned/gpt'd the
disk before trying to reconstruct the raid to it?

2.  Is the new drive really identical, or is it a similarly sized drive?
I've seendrives that purport to be the same size but have different
absolute sector counts, differing by as much as a few hundred sectors.
If the new drive's raid partition, however you've defined it, is smaller
than the original raid partition, even as little as one sector, that would
cause the behavior you describe.

3.  Is it possible you're running into some protected sectors or something
like that on the new drive?  Was the drive previously used elsewhere?


-thanks
-Brian

On Mar 6, 12:51pm, Greg Oster wrote:
} Subject: Re: NetBSD RAIDFrame issue (kind of off-topic).
} On Tue, 06 Mar 2012 12:34:42 -0600
} Dave Burgess <burgess%cynjut.com@localhost> wrote:
} 
} > Guys,
} > 
} > I'm having a weird RAID-5 problem and I'd like to chat with someone 
} > about it.  This isn't really a NetBSD issue, even though the RAID
} > array is the NetBSD RaidFrame.  You all have more important stuff to
} > deal with.  If someone could pop back to be, I'd appreciate it.
} > 
} > Here's the issue:
} > 
} > I recently had a drive in a 4 drive raid-5 array fail.  I pulled the 
} > drive (bearing problems - wouldn't even turn) and replaced it with an 
} > identical drive (in geometry).
} > 
} > I issued a raidctl -R /dev/wd3a raid0 on the array to rebuild the new 
} > drive.  It runs for about an hour and dies at the last second with an 
} > error (which I don't have right handy - the drive is at a customer's
} > site).
} > 
} > When I try to mount the array, one half of the array (/dev/raid0e)
} > works fine, but the second half (/dev/raid0f) won't mount with an
} > error "No space left on device".  It is possible that the drive is
} > over full, but it seems to me that it's about 70%.
} > 
} > The /var partition is actually full, so there might be an issue with
} > a log message getting jammed, but would that kill the raidctl -R at
} > the last second?
} > 
} > Any suggestions?
} 
} Really need to see the error (and whatever's relevant from dmesg)....  
} My guess is that you're hitting a read error on one of the other
} drives -- that should be about the only thing that would cause the
} 'raidctl -R' to fail... (or a write error to the new drive...)
} 
} Later...
} 
} Greg Oster
>-- End of excerpt from Greg Oster