On Fri, Jan 11, 2013 at 09:51:22AM +1100, Dave Chinner wrote:
> On Fri, Jan 11, 2013 at 12:34:24AM +0800, Eryu Guan wrote:
> > Run 1000 fsstress processes and freeze/unfreeze file system randomly.
> > Signed-off-by: Eryu Guan <eguan@xxxxxxxxxx>
> > ---
> > 293 | 88
> > +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > 293.out | 4 +++
> > group | 1 +
> > 3 files changed, 93 insertions(+)
> > create mode 100755 293
> > create mode 100644 293.out
> > diff --git a/293 b/293
> > new file mode 100755
> > index 0000000..a360e9c
> > --- /dev/null
> > +++ b/293
> > @@ -0,0 +1,88 @@
> > +#! /bin/bash
> > +# FS QA Test No. 293
> > +#
> > +# Test freeze/unfreeze file system randomly under fsstress
> What does this test do differently to 068? i.e. why do we need two
> generic fsstress vs freeze tests?
068 does fixed number of iterations and sleep 2 seconds between each
iteration and freeze/unfreeze operation. Also 068 only forks 2 fsstress
processes, the load is relatively low.
This test forks 1000 fsstress processes and does random freeze/unfreeze
operations until all fsstress processes have done their jobs.
Basically it's low stress with fixed ops vs relatively high stress with
I agree that this test is a bit redundant, but I still perfer sending it
out for review, if it's not needed I can drop it in next version.
> Dave Chinner