On Tue, May 29, 2018 at 09:30:23AM -0700, Darrick J. Wong wrote:
> On Wed, May 30, 2018 at 12:13:04AM +0800, Zorro Lang wrote:
> > On Tue, May 29, 2018 at 08:07:59AM -0700, Darrick J. Wong wrote:
> > > On Mon, May 28, 2018 at 12:54:27PM +0800, Zorro Lang wrote:
> > > > Duperemove is a tool for finding duplicated extents and submitting
> > > > them for deduplication, and it supports XFS. This case trys to
> > > > verify the integrity of XFS after running duperemove.
> > > >
> > > > Signed-off-by: Zorro Lang <zlang@xxxxxxxxxx>
> > > > ---
> > > >
> > > > Hi,
> > > >
> > > > There's not many softwares support XFS dedup now, duperemove is a rare one.
> > > > So I write this case by using duperemove.
> > > >
> > > > I use fsstress to make many files and data randomly, I don't know if there're
> > > > better things I can use? Because fsstress only write '0xff' into files, maybe
> > > > I should add an option to make fsstress can write random character?
> > >
> > > Heh. But you probably don't want totally random contents because then
> > > dupremove doesn't do much.
> >
> > No matter how random contents I get, I will copy once :)
>
> I suppose so. Once fsstress pokes reflink enough there ought to be a
> fair number of easy targets for dedupe... on the other hand I think it's
> a useful test for "literally everything in this fs is identical, dedupe
> everything" :)
Do you mean dd (with same char) a big file to fill the whole scratch_dev, then
dedupe?
Hmm... how about iteration test as below?
# fsstress -d $mnt/dir0 ...
# for ((i=1; i<100; i++));do
cp -a dir$((i-1)) dir$i
find $dir$i -type f -exec md5sum {} \; > $TEST_DIR/${seq}md5.sum$i
duperemove -dr $mnt/
md5sum -c $TEST_DIR/${seq}md5.sum$i
done
# scratch_cycle_mount
# for ((i=1; i<100; i++));do
md5sum -c $TEST_DIR/${seq}md5.sum$i
done
But this will cost lots of test time. For save time, need to reduce files count
and size.
Hmm... if we have different dedupe test ways, maybe we can write not only one
cases.
>
> > >
> > > >
> > > > Please tell me, if you have better ideas:)
> > > >
> > > > PS: This case test passed on XFS(with reflink=1) and btrfs. And the duperemove
> > > > can reclaim some space in the test, see below:
> > > >
> > > > Before duperemove
> > > > Filesystem 1K-blocks Used Available Use% Mounted on
> > > > /dev/mapper/xxxx-xfscratch 31441920K 583692K 30858228K 2% /mnt/scratch
> > > >
> > > > After duperemove
> > > > Filesystem 1K-blocks Used Available Use% Mounted on
> > > > /dev/mapper/xxxx-xfscratch 31441920K 345728K 31096192K 2% /mnt/scratch
> > > >
> > > > Thanks,
> > > > Zorro
> > > >
> > > > common/config | 1 +
> > > > tests/shared/008 | 88 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > > tests/shared/008.out | 2 ++
> > > > tests/shared/group | 1 +
> > > > 4 files changed, 92 insertions(+)
> > > > create mode 100755 tests/shared/008
> > > > create mode 100644 tests/shared/008.out
> > > >
> > > > diff --git a/common/config b/common/config
> > > > index 02c378a9..def559c1 100644
> > > > --- a/common/config
> > > > +++ b/common/config
> > > > @@ -207,6 +207,7 @@ export SQLITE3_PROG="`set_prog_path sqlite3`"
> > > > export TIMEOUT_PROG="`set_prog_path timeout`"
> > > > export SETCAP_PROG="`set_prog_path setcap`"
> > > > export GETCAP_PROG="`set_prog_path getcap`"
> > > > +export DUPEREMOVE_PROG="`set_prog_path duperemove`"
> > > >
> > > > # use 'udevadm settle' or 'udevsettle' to wait for lv to be settled.
> > > > # newer systems have udevadm command but older systems like RHEL5 don't.
> > > > diff --git a/tests/shared/008 b/tests/shared/008
> > > > new file mode 100755
> > > > index 00000000..dace5429
> > > > --- /dev/null
> > > > +++ b/tests/shared/008
> > > > @@ -0,0 +1,88 @@
> > > > +#! /bin/bash
> > > > +# FS QA Test 008
> > > > +#
> > > > +# Dedup integrity test by duperemove
> > > > +#
> > > > +#-----------------------------------------------------------------------
> > > > +# Copyright (c) 2018 Red Hat Inc. 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
> > > > +. ./common/reflink
> > > > +
> > > > +# remove previous $seqres.full before test
> > > > +rm -f $seqres.full
> > > > +
> > > > +# real QA test starts here
> > > > +
> > > > +# duperemove only supports btrfs and xfs (with reflink feature).
> > > > +# Add other filesystems if it supports more later.
> > > > +_supported_fs xfs btrfs
> > > > +_supported_os Linux
> > >
> > > _require_command "$DUPEREMOVE_PROG" duperemove ?
> >
> > Yes, it would be better to use this template, not check
> > [ "$DUPEREMOVE_PROG" = "" ].
> >
> > >
> > > > +_require_scratch_reflink
> > >
> > > _require_scratch_dedupe
> >
> > Yes, I should check XFS_IOC_FILE_EXTENT_SAME, not XFS_IOC_CLONE*.
> >
> > >
> > > > +
> > > > +[ "$DUPEREMOVE_PROG" = "" ] && _notrun "duperemove not found"
> > > > +_scratch_mkfs > $seqres.full 2>&1
> > > > +_scratch_mount >> $seqres.full 2>&1
> > > > +
> > > > +testdir=$SCRATCH_MNT/test-$seq
> > > > +mkdir $testdir
> > > > +
> > > > +fsstress_opts="-w -r -f mknod=0"
> > > > +# Create some files and make a duplicate
> > > > +$FSSTRESS_PROG $fsstress_opts -d $testdir \
> > > > + -n $((500 * LOAD_FACTOR)) -p 10 >/dev/null 2>&1
> > > > +duptestdir=${testdir}.dup
> > > > +cp -a $testdir $duptestdir
> > > > +
> > > > +# Make some difference in two directories
> > > > +$FSSTRESS_PROG $fsstress_opts -d $testdir -n 200 -p 5 >/dev/null 2>&1
> > > > +$FSSTRESS_PROG $fsstress_opts -d $duptestdir -n 200 -p 5 >/dev/null 2>&1
> > > > +
> > > > +# Record all files' md5 checksum
> > > > +find $testdir -type f -exec md5sum {} \; > $TEST_DIR/${seq}md5.sum
> > > > +find $duptestdir -type f -exec md5sum {} \; > $TEST_DIR/dup${seq}md5.sum
> > > > +
> > > > +# Dedup
> > > > +echo "== Duperemove output ==" >> $seqres.full
> > > > +$DUPEREMOVE_PROG -dr $SCRATCH_MNT/ >>$seqres.full 2>&1
> > > > +
> > > > +# Verify all files' integrity
> > > > +md5sum -c --quiet $TEST_DIR/${seq}md5.sum
> > > > +md5sum -c --quiet $TEST_DIR/dup${seq}md5.sum
> > >
> > > Can we _scratch_mount_cycle and md5sum -c again so that we test that the
> > > pagecache contents don't mutate and a fresh read from the disk also
> > > doesn't show mutations?
> >
> > If so, is the md5sum data safe? Should I do cycle_mount before get md5 checksum?
> > What 'fresh read' do you mean, from above duperemove processes? Or you hope to
> > read all files once before cycle_mount?
>
> Since this is dedupe, the md5sum should never change. The existing
> md5sum -c check makes sure that the dedupe operation doesn't
> mishandle/corrupt the page cache such that it suddenly starts returning
> incorrect contents; and the post-cycle md5sum -c check I propose would
> flush the page cache and make sure that the on-disk contents also have
> not changed.
Make sense, I'll do a cycle_mount and md5sum -c check again.
>
> --D
>
> > Thanks,
> > Zorro
> >
> > >
> > > --D
> > >
> > > > +
> > > > +echo "Silence is golden"
> > > > +
> > > > +status=0
> > > > +exit
> > > > diff --git a/tests/shared/008.out b/tests/shared/008.out
> > > > new file mode 100644
> > > > index 00000000..dd68d5a4
> > > > --- /dev/null
> > > > +++ b/tests/shared/008.out
> > > > @@ -0,0 +1,2 @@
> > > > +QA output created by 008
> > > > +Silence is golden
> > > > diff --git a/tests/shared/group b/tests/shared/group
> > > > index b3663a03..de7fe79f 100644
> > > > --- a/tests/shared/group
> > > > +++ b/tests/shared/group
> > > > @@ -10,6 +10,7 @@
> > > > 005 dangerous_fuzzers
> > > > 006 auto enospc
> > > > 007 dangerous_fuzzers
> > > > +008 auto quick dedupe
> > > > 032 mkfs auto quick
> > > > 272 auto enospc rw
> > > > 289 auto quick
> > > > --
> > > > 2.14.3
> > > >
> > > > --
> > > > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > > > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > > > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> --
> To unsubscribe from this list: send the line "unsubscribe fstests" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
