Re: [PATCH] generic: create and delete files repeatedly to exercise ENOSPC behaviour

[Eryu Guan <eguan@redhat.com>]

On Tue, Nov 01, 2016 at 07:19:30PM +0800, Wang Xiaoguang wrote:
> In btrfs, sometimes though the number of created files' consumed disk space
> are not larger than fs's free space, we can still get some ENOSPC error, it
> may be that btrfs does not try hard to reclaim disk space(I have sent kernel
> patch to resolve this kind of enospc error. Note, this false enospc error
> will not always happen even in kernel without my fixing patch).
> 
> Currently only in btrfs, I get this ENOSPC error, xfs and ext4 work well.
> 
> Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com>
> ---
>  tests/generic/389     | 78 +++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/generic/389.out |  2 ++
>  tests/generic/group   |  1 +
>  3 files changed, 81 insertions(+)
>  create mode 100755 tests/generic/389
>  create mode 100644 tests/generic/389.out
> 
> diff --git a/tests/generic/389 b/tests/generic/389
> new file mode 100755
> index 0000000..96bc12e
> --- /dev/null
> +++ b/tests/generic/389
> @@ -0,0 +1,78 @@
> +#! /bin/bash
> +# FS QA Test 389
> +#
> +# Create and delete files repeatedly to exercise ENOSPC behaviour. 

Trailing whitespace in this line.

> +#
> +#-----------------------------------------------------------------------
> +# Copyright (c) 2016 Fujitsu.  All Rights Reserved.
> +#
> +# This program is free software; you can redistribute it and/or
> +# modify it under the terms of the GNU General Public License as
> +# published by the Free Software Foundation.
> +#
> +# This program is distributed in the hope that it would be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program; if not, write the Free Software Foundation,
> +# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
> +#-----------------------------------------------------------------------
> +#
> +
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1	# failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> +	cd /
> +	rm -f $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# Modify as appropriate.
> +_supported_fs generic
> +_supported_os Linux
> +_require_scratch
> +
> +RUN_TIME=$((600 * $TIME_FACTOR))

Hmm, does it really need 600s to run? I think it's better to limit the
runtime within 300s and make it an 'auto' test. I, personally, prefer a
"loop count" based test, I'd find out a minimum loop count that could
reproduce the ENOSPC problem more reliably on btrfs (for example, say
75%) and make the count scale with LOAD_FACTOR.

> +fs_size=$((15 * 1024 * 1024 * 1024))

And does it really need 15G on SCRATCH_DEV? A smaller fs size makes test
run faster, and gives the test more chance to be run, because not
everyone has a 15G SCRATCH_DEV.

> +_scratch_mkfs_sized $fs_size > $seqres.full 2>&1
> +_scratch_mount > $seqres.full 2>&1

Append to $seqres.full not overwrite.

> +
> +testfile1=$SCRATCH_MNT/testfile1
> +testfile2=$SCRATCH_MNT/testfile2
> +filesize1=$(((fs_size * 80) / 100))
> +filesize2=$(((fs_size * 5) / 100))

Better to have some comments on the filesizes chosen here. e.g. someone
may wonder that why it's testing ENOSPC condition with 85% full, not 99%
or 100%.

> +
> +do_test()
> +{
> +	while [ -f $SCRATCH_MNT/run ]; do
> +		$XFS_IO_PROG -fc "pwrite 0 $filesize1" $testfile1 > /dev/null
> +		$XFS_IO_PROG -fc "pwrite 0 $filesize2" $testfile2 > /dev/null
> +		rm -f $testfile1 $testfile2 

Trailing whitespace here.

> +	done
> +}
> +
> +echo "Silence is golden"
> +touch $SCRATCH_MNT/run
> +do_test &
> +sleep $RUN_TIME
> +rm -f $SCRATCH_MNT/run
> +wait
> +
> +status=0
> +exit
> diff --git a/tests/generic/389.out b/tests/generic/389.out
> new file mode 100644
> index 0000000..e8c24bb
> --- /dev/null
> +++ b/tests/generic/389.out
> @@ -0,0 +1,2 @@
> +QA output created by 389
> +Silence is golden

Can you please rebase on top of current master? generic/389 is already
taken, and it makes applying & testing the patch a litter harder :)

> diff --git a/tests/generic/group b/tests/generic/group
> index fc32cfd..b6d4013 100644
> --- a/tests/generic/group
> +++ b/tests/generic/group
> @@ -391,3 +391,4 @@
>  386 auto quick quota
>  387 auto clone
>  388 auto log metadata
> +389 enospc

Perhaps we can add it to 'rw' group too.

Thanks,
Eryu