PERFORCE change 121535 for review

[Ulf Lilleengen]

On Fri, Jun 15, 2007 at 02:49:40PM +0200, Pawel Jakub Dawidek wrote:
> On Tue, Jun 12, 2007 at 08:57:20PM +0000, Ulf Lilleengen wrote:
> > http://perforce.freebsd.org/chv.cgi?CH=121535
> > 
> > Change 121535 by lulf at lulf_carrot on 2007/06/12 20:57:13
> > 
> > 	- Allow subdisks to be added to raid5 plexes again.
> > 	- Add detach to userland gvinum tool
> > 	- Add gv_detach general request handler for detaching
> > 	- Add gv_detach_sd and gv_detach_plex to detach a subdisk and a plex.
> > 	- Make gv_sd_to_plex use the standard plex-size calculation.
> > 	- Add a sddetached-variable to gv_plex do be able to detect wether a
> > 	  plex misses a subdisk or not. This variable is increased when a
> > 	  subdisk is detached, and increased when a subdisk is attached.
> > 	- Remove weird offset-check with a correct one.
> > 	- Hook it up in the event system.
> > 	- Looks like handling a crashed drive, and replacing it is working so
> > 	  far!  
> 
> What do you use to test this? If you're not using it now, you should try
> gnop(8). You can simulate disk removal and I/O errors. It proved to be
> very handy for me in the past.
> 
I do a dd if=/dev/zero of=/dev/<gvinumdrive-device> which makes gvinum "loose"
the drive. But thanks for the tip. It can perhaps give me ability to test more
than just removal.

-- 
Ulf Lilleengen