Re: kern/38856: Raidframe lockups

[Greg Oster <oster%cs.usask.ca@localhost>]

The following reply was made to PR kern/38856; it has been noted by GNATS.

From: Greg Oster <oster%cs.usask.ca@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/38856: Raidframe lockups 
Date: Thu, 05 Jun 2008 08:55:37 -0600

 Martin Husemann writes:
 > The following reply was made to PR kern/38856; it has been noted by GNATS.
 > 
 > From: Martin Husemann <martin%duskware.de@localhost>
 > To: gnats-bugs%NetBSD.org@localhost
 > Cc: 
 > Subject: Re: kern/38856: Raidframe lockups
 > Date: Thu, 5 Jun 2008 09:40:52 +0200
 > 
 >  I had to do more parity rewrites on the other raid, and saw the lockup there
 >  again too. Maybe it is related to the amount of things written and since
 >  that raid-1 mostly was fine the parity rebuild rarely causes it, but since
 >  the other raid-1 has a complete fresh empty disk, the reconstruction has
 >  to write every single block and thus triggers it more easily?
 >  
 >  Christos says he can also reproduce it by writes from userland via the
 >  filesystem (on the raid), but I haven't been able to reproduce that.
 >  
 >  Reading from the raid in userland and writing to another (non-raid) disk
 >  works just fine (both via dd on the raw disk or via the filesystem on raid)
 
 I've now been able to reproduce this...  doing a 'raidctl -i raid2' 
 while 5 of these:  'dd if=/dev/zero of=bigfile_n bs=10m count=1000' 
 are running is what did it for me...
 
 Later...
 
 Greg Oster