On Fri, May 13, 2005 at 01:15:16PM +0200, Jukka Salmi wrote:
> Hi,
> 
> I'm using RAIDframe (level 1) on a 3.99.3 i386 system with two IDE disks.
> Becuase of a problem I was having with one of the drives, and decided to
> switch the drives. Luckily I had a spare, so I could do this by
> 
> 	- replacing drive A (original wd0) with the spare
> 	- reconstructing onto the spare
> 	- replacing drive B (original wd1) with drive A
> 	- reconstructin onto drive A
> 	- replacing the spare with drive B
> 	- reconstructing onto drive B
> 
> The toughest part of this process was to not mix the drives while switching
> them around...
> 
> Eventually everything went fine; however, while reconstructing the disks
> I noticed a problem: the first time I reconstructed, the progress bar
> stalled at 29% and ETA constantly increased until reconstruction eventually
> succeeded. During the other two reconstructions, the progress bar always
> printed 0% and ETA 0:00 until it completed. However, reconstruction was
> always successful (AFAICT), and never took longer than expected.
> 
> What could be the source if this problem?

There were some 32-bit overflow problems with raidctl which Greg fixed
recently - I saw some 64 bit values were being returned properly, but
were being stored in 32 bit values.

	http://mail-index.netbsd.org/source-changes/2005/05/01/0122.html

(it's time for me to squeeze out of the closet and own up to being
"beefy").

Regards,
Alistair