Re: [PATCH v2 5/6] fstests: btrfs: test RAID1 device reappear and balanced

[Eryu Guan <eguan@redhat.com>]

On Thu, Jun 30, 2016 at 07:04:06PM +0800, Anand Jain wrote:
> 
> 
> Thanks for review comments.
> more below..
> 
> On 06/27/2016 05:29 PM, Eryu Guan wrote:
> > On Wed, Jun 22, 2016 at 07:01:54PM +0800, Anand Jain wrote:
> > > 
> > > 
> > > On 06/21/2016 09:31 PM, Eryu Guan wrote:
> > > > On Wed, Jun 15, 2016 at 04:48:47PM +0800, Anand Jain wrote:
> > > > > From: Anand Jain <Anand.Jain@oracle.com>
> > > > > 
> > > > > The test does the following:
> > > > >   Initialize a RAID1 with some data
> > > > > 
> > > > >   Re-mount RAID1 degraded with _dev1_ and write up to
> > > > >   half of the FS capacity
> > > > 
> > > > If test devices are big enough, this test consumes much longer test
> > > > time. I tested with 15G scratch dev pool and this test ran ~200s on my
> > > > 4vcpu 8G memory test vm.
> > > 
> > >  Right. Isn't that a good design? So that it gets tested differently
> > >  on different HW config. ?
> > 
> > Not in fstests. We should limit the run time of tests to an acceptable
> > amount, for auto group it's within 5 minutes.
> 
> > >  However the test time can be reduced by using smaller vdisk.
> > 
> > I think either limit the write size or _notrun if the $max_fs_size is
> > too big (say 30G).
> 
> Fixed in v3 to have a fixed scratch data.

Thanks! I've queued this patchset up, will let them go through some
testings.

Thanks,
Eryu