Subject: Re: kern/35580
To: None <kern-bug-people@netbsd.org, gnats-admin@netbsd.org,>
From: Jeff Rizzo <riz@NetBSD.org>
List: netbsd-bugs
Date: 03/02/2007 21:00:09
The following reply was made to PR kern/35580; it has been noted by GNATS.

From: Jeff Rizzo <riz@NetBSD.org>
To: gnats-bugs@netbsd.org
Cc: 
Subject: Re: kern/35580
Date: Fri, 2 Mar 2007 11:53:41 -0800

 More investigation has suggested to me that this is a file system
 problem rather than a controller problem.
 
 Using the above test (creating 1000 directories - I think the files
 within the directories are irrelevant, though I haven't tested without
 them), I get a repeatable failure with a filesystem of 8T or above.
 I haven't narrowed much within the range, but creating progressively
 larger and smaller arrays with the twa(4) controller, using 750G drives,
 an 8.1T file system (13 drives in a raid5) exhibits the problem, while
 a 7.4T file system does not.  An 8.9T file system exhibits the problem.
 
 Also, taking the initial, problematic 8.1T array and putting two 4T file
 systems on it yields two _working_ file systems.
 
 It seems pretty clear to me that we have some sort of overflow issue
 going on.