From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from cn.fujitsu.com ([59.151.112.132]:56365 "EHLO heian.cn.fujitsu.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1753137AbcKBK3i (ORCPT ); Wed, 2 Nov 2016 06:29:38 -0400 Subject: Re: [PATCH] generic: create and delete files repeatedly to exercise ENOSPC behaviour References: <1477999170-20026-1-git-send-email-wangxg.fnst@cn.fujitsu.com> <20161101122613.GR27776@eguan.usersys.redhat.com> From: Wang Xiaoguang Message-ID: <5819BE82.10205@cn.fujitsu.com> Date: Wed, 2 Nov 2016 18:22:58 +0800 MIME-Version: 1.0 In-Reply-To: <20161101122613.GR27776@eguan.usersys.redhat.com> Content-Type: text/plain; charset="windows-1252"; format=flowed Content-Transfer-Encoding: 7bit Sender: fstests-owner@vger.kernel.org To: Eryu Guan Cc: fstests@vger.kernel.org, linux-btrfs@vger.kernel.org List-ID: hi Eryu, There has already be a generic/102 doing this test... Thanks for you kindly review and sorry for wasting your time. Regards, Xiaoguang Wang On 11/01/2016 08:26 PM, Eryu Guan wrote: > 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 >> --- >> 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 > >