Re: [PATCH v2 3/3] xfstests: dedupe with random io race test

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Wed, Jun 20, 2018 at 04:41:14PM +0800, Zorro Lang wrote:
> Run several duperemove processes with fsstress on same directory at
> same time. Make sure the race won't break the fs or kernel.
> 
> Signed-off-by: Zorro Lang <zlang@redhat.com>
> ---
> 
> V2 did below changes:
> 1) do sleep 1 after kill processes
> 2) change SLEEP_TIME to sleep_time
> 3) add the case to stress group
> 
> Thanks,
> Zorro
> 
>  tests/shared/010     | 111 +++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/shared/010.out |   2 +
>  tests/shared/group   |   1 +
>  3 files changed, 114 insertions(+)
>  create mode 100755 tests/shared/010
>  create mode 100644 tests/shared/010.out
> 
> diff --git a/tests/shared/010 b/tests/shared/010
> new file mode 100755
> index 00000000..c449c247
> --- /dev/null
> +++ b/tests/shared/010
> @@ -0,0 +1,111 @@
> +#! /bin/bash
> +# FS QA Test 010
> +#
> +# Dedup & random I/O race test, do multi-threads fsstress and dedupe on
> +# same directory/files
> +#
> +#-----------------------------------------------------------------------
> +# 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.*
> +	kill_all_stress
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +. ./common/reflink
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +# duperemove only supports btrfs and xfs (with reflink feature).
> +# Add other filesystems if it supports more later.
> +_supported_fs xfs btrfs
> +_supported_os Linux
> +_require_scratch_dedupe
> +_require_command "$DUPEREMOVE_PROG" duperemove
> +_require_command "$KILLALL_PROG" killall
> +
> +_scratch_mkfs > $seqres.full 2>&1
> +_scratch_mount >> $seqres.full 2>&1
> +
> +function kill_all_stress()
> +{
> +	local f=1
> +	local d=1
> +
> +	# kill the bash process which loop run duperemove
> +	if [ -n "$loop_dedup_pid" ]; then
> +		kill $loop_dedup_pid > /dev/null 2>&1
> +		wait $loop_dedup_pid > /dev/null 2>&1
> +		loop_dedup_pid=""
> +	fi
> +
> +	# Make sure all fsstress and duperemove processes get killed
> +	while [ $((f + d)) -ne 0 ]; do
> +		$KILLALL_PROG -q $FSSTRESS_PROG > /dev/null 2>&1
> +		$KILLALL_PROG -q $DUPEREMOVE_PROG > /dev/null 2>&1
> +		sleep 1
> +		f=`ps -eLf | grep $FSSTRESS_PROG | grep -v "grep" | wc -l`
> +		d=`ps -eLf | grep $DUPEREMOVE_PROG | grep -v "grep" | wc -l`
> +	done
> +}
> +
> +sleep_time=$((50 * TIME_FACTOR))
> +
> +# Start fsstress
> +fsstress_opts="-r -n 1000 -p $((5 * LOAD_FACTOR))"
> +$FSSTRESS_PROG $fsstress_opts -d $SCRATCH_MNT -l 0 >> $seqres.full 2>&1 &
> +loop_dedup_pid=""
> +# Start several dedupe processes on same directory
> +for ((i = 0; i < $((2 * LOAD_FACTOR)); i++)); do
> +	while true; do
> +		$DUPEREMOVE_PROG -dr --dedupe-options=same $SCRATCH_MNT/ \
> +			>>$seqres.full 2>&1

/me wonders why not just touch $TEST_DIR/run, have this loop do:

while test -e $TEST_DIR/run; do
	duperemove...
done

and then rm -f $TEST_DIR/run to end the loop?  Then you don't need the
complex machinery to shut down the bash loop and kill the fsstress and
duperemove processes.

Otherwise looks decent,

--D

> +	done &
> +	loop_dedup_pid="$! $loop_dedup_pid"
> +done
> +
> +# End the test after $sleep_time seconds
> +sleep $sleep_time
> +kill_all_stress
> +
> +# umount and mount again, verify pagecache contents don't mutate and a fresh
> +# read from the disk also doesn't show mutations.
> +find $testdir -type f -exec md5sum {} \; > $TEST_DIR/${seq}md5.sum
> +_scratch_cycle_mount
> +md5sum -c --quiet $TEST_DIR/${seq}md5.sum
> +
> +echo "Silence is golden"
> +status=0
> +exit
> diff --git a/tests/shared/010.out b/tests/shared/010.out
> new file mode 100644
> index 00000000..1d83a8d6
> --- /dev/null
> +++ b/tests/shared/010.out
> @@ -0,0 +1,2 @@
> +QA output created by 010
> +Silence is golden
> diff --git a/tests/shared/group b/tests/shared/group
> index 9c484794..094da27d 100644
> --- a/tests/shared/group
> +++ b/tests/shared/group
> @@ -12,6 +12,7 @@
>  007 dangerous_fuzzers
>  008 auto stress dedupe
>  009 auto stress dedupe
> +010 auto stress dedupe
>  032 mkfs auto quick
>  272 auto enospc rw
>  289 auto quick
> -- 
> 2.14.4
> 
> --
> To unsubscribe from this list: send the line "unsubscribe fstests" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html