[sheepdog] [PATCH v2 1/2] tests/func: refine 072

[MORITA Kazutaka]

At Mon, 2 Sep 2013 13:28:50 +0800,
Liu Yuan wrote:
> 
> On Mon, Sep 02, 2013 at 02:14:29PM +0900, MORITA Kazutaka wrote:
> > At Fri, 30 Aug 2013 22:56:32 +0800,
> > Liu Yuan wrote:
> > > 
> > > We should use '_kill_all_sheep' instead of 'cluster shutdown' to simulate a
> > > cluster crash
> > > 
> > > Signed-off-by: Liu Yuan <namei.unix at gmail.com>
> > > ---
> > >  tests/functional/072 |    2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/tests/functional/072 b/tests/functional/072
> > > index 4f7a7c3..47086df 100755
> > > --- a/tests/functional/072
> > > +++ b/tests/functional/072
> > > @@ -17,7 +17,7 @@ _start_sheep 2
> > >  _wait_for_sheep 3
> > >  
> > >  $DOG vdi create test 8M -P
> > > -$DOG cluster shutdown
> > > +_kill_all_sheeps
> > >  
> > >  mv $STORE/0/obj/007c2b2500000001 $STORE/0/obj/.stale/007c2b2500000001.1
> > >  mv $STORE/1/obj/807c2b2500000000 $STORE/1/obj/.stale/807c2b2500000000.1
> > 
> > If we use _kill_all_sheeps(), $STORE/0/obj/007c2b2500000001 and
> > $STORE/1/obj/807c2b2500000000 may not exist, no?
> 
> Why? _kill_all_sheeps kill the sheep in the same time, why may not exist?

I don't think the command kill the sheep at the same time.  It calls
killall, which will send a signal to the matched processes
sequentially.

Thanks,

Kazutaka