The following reply was made to PR kern/29881; it has been noted by GNATS.

From: Denis Lagno <dlagno@rambler.ru>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/29881: umount stale NFS volume blocks deadly
Date: Tue, 5 Apr 2005 10:16:02 +0400

 On Mon, Apr 04, 2005 at 04:31:53PM -0500, Frederick Bruckman scribed:
 > In article <20050404193608.GC1257@chup.gado>,
 > 	Denis Lagno <dlagno@rambler.ru> writes:
 > > On Mon, Apr 04, 2005 at 07:12:02PM +0000, Manuel Bouyer scribed:
 > >>  > I mount NFS volume, then for some reason I lose the network.  Then umount this volume just blocks.
 > >>  > So if I reboot, it stops in "Unmounting file systems" and no progress ever seen.
 > >>  
 > >>  This is the proper behavior. Unless mounted -o soft, the client has to wait for
 > >>  the server to come back.
 > 
 > Or even better, mount with "-i", interruptible. Then you can choose to
 > let the client wait patiently, forever, for the server to come back up,
 > or you can kill any processes waiting to write and forcibly unmount.
 > 
 > > Then probably shutdown script should remount NFS volumes as soft, and then umount them.
 > > Or something like that..
 > > Anyway it seems weird that umount -f cannot forcibly umount filesystem.
 > 
 > If you really don't want the last bit of data to be written out, then
 > why do you care about a clean shutdown?  You know what you have to do.
 > Mounting with "-i", though, mostly does what you want -- on shutdown,
 > it'll let the local disks sync, while trashing the data waiting to go
 > over the network.
 
 OK, I see.  Then I do not object closing PR.