* [PATCH] fstest: random read fio test for read policy
@ 2021-02-22 14:48 Anand Jain
2021-03-07 17:54 ` Eryu Guan
2021-03-10 23:40 ` Anand Jain
0 siblings, 2 replies; 3+ messages in thread
From: Anand Jain @ 2021-02-22 14:48 UTC (permalink / raw)
To: fstests; +Cc: linux-btrfs
This test case runs fio for raid1/10/1c3/1c4 profiles and all the
available read policies in the system. At the end of the test case,
a comparative summary of the result is in the $seqresfull-file.
LOAD_FACTOR parameter controls the fio scalability. For the
LOAD_FACTOR = 1 (default), this runs fio for file size = 1G and num
of jobs = 1, which approximately takes 65s to finish.
There are two objectives of this test case. 1. by default, with
LOAD_FACTOR = 1, it sanity tests the read policies. And 2. Run the
test case individually with a larger LOAD_FACTOR. For example, 10
for the comparative study of the read policy performance.
I find tests/btrfs as the placeholder for this test case. As it
contains many things which are btrfs specific and didn't fit well
under perf.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
tests/btrfs/231 | 145 ++++++++++++++++++++++++++++++++++++++++++++
tests/btrfs/231.out | 2 +
tests/btrfs/group | 1 +
3 files changed, 148 insertions(+)
create mode 100755 tests/btrfs/231
create mode 100644 tests/btrfs/231.out
diff --git a/tests/btrfs/231 b/tests/btrfs/231
new file mode 100755
index 000000000000..c08b5826f60a
--- /dev/null
+++ b/tests/btrfs/231
@@ -0,0 +1,145 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2021 Anand Jain. All Rights Reserved.
+#
+# FS QA Test 231
+#
+# Random read fio test for raid1(10)(c3)(c4) with available
+# read policy.
+#
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+fio_config=$tmp.fio
+fio_results=$tmp.fio_out
+
+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 $seqres.full
+
+# real QA test starts here
+
+# Modify as appropriate.
+_supported_fs btrfs
+_require_scratch_dev_pool 4
+
+njob=$LOAD_FACTOR
+size=$LOAD_FACTOR
+_require_scratch_size $(($size * 2 * 1024 * 1024))
+echo size=$size njob=$njob >> $seqres.full
+
+make_fio_config()
+{
+ #Set direct IO to true, help to avoid buffered IO so that read happens
+ #from the devices.
+ cat >$fio_config <<EOF
+[global]
+bs=64K
+iodepth=64
+direct=1
+invalidate=1
+allrandrepeat=1
+ioengine=libaio
+group_reporting
+size=${size}G
+rw=randread
+EOF
+}
+#time_based
+#runtime=5
+
+make_fio_config
+for job in $(seq 0 $njob); do
+ echo "[foo$job]" >> $fio_config
+ echo "filename=$SCRATCH_MNT/$job/file" >> $fio_config
+done
+_require_fio $fio_config
+cat $fio_config >> $seqres.full
+
+work()
+{
+ raid=$1
+
+ echo ------------- profile: $raid ---------- >> $seqres.full
+ echo >> $seqres.full
+ _scratch_pool_mkfs $raid >> $seqres.full 2>&1
+ _scratch_mount
+
+ fsid=$($BTRFS_UTIL_PROG filesystem show -m $SCRATCH_MNT | grep uuid: | \
+ $AWK_PROG '{print $4}')
+ readpolicy_path="/sys/fs/btrfs/$fsid/read_policy"
+ policies=$(cat $readpolicy_path | sed 's/\[//g' | sed 's/\]//g')
+
+ for policy in $policies; do
+ echo $policy > $readpolicy_path || _fail "Fail to set readpolicy"
+ echo -n "activating readpolicy: " >> $seqres.full
+ cat $readpolicy_path >> $seqres.full
+ echo >> $seqres.full
+
+ > $fio_results
+ $FIO_PROG --output=$fio_results $fio_config
+ cat $fio_results >> $seqres.full
+ done
+
+ _scratch_unmount
+ _scratch_dev_pool_put
+}
+
+_scratch_dev_pool_get 2
+work "-m raid1 -d single"
+
+_scratch_dev_pool_get 2
+work "-m raid1 -d raid1"
+
+_scratch_dev_pool_get 4
+work "-m raid10 -d raid10"
+
+_scratch_dev_pool_get 3
+work "-m raid1c3 -d raid1c3"
+
+_scratch_dev_pool_get 4
+work "-m raid1c4 -d raid1c4"
+
+
+# Now benchmark the raw device performance
+> $fio_config
+make_fio_config
+_scratch_dev_pool_get 4
+for dev in $SCRATCH_DEV_POOL; do
+ echo "[$dev]" >> $fio_config
+ echo "filename=$dev" >> $fio_config
+done
+_require_fio $fio_config
+cat $fio_config >> $seqres.full
+
+echo ------------- profile: raw disk ---------- >> $seqres.full
+echo >> $seqres.full
+> $fio_results
+$FIO_PROG --output=$fio_results $fio_config
+cat $fio_results >> $seqres.full
+
+echo >> $seqres.full
+echo ===== Summary ====== >> $seqres.full
+cat $seqres.full | egrep -A1 "Run status|Disk stats|profile:|readpolicy" >> $seqres.full
+
+echo "Silence is golden"
+
+# success, all done
+status=0
+exit
diff --git a/tests/btrfs/231.out b/tests/btrfs/231.out
new file mode 100644
index 000000000000..a31b87a289bf
--- /dev/null
+++ b/tests/btrfs/231.out
@@ -0,0 +1,2 @@
+QA output created by 231
+Silence is golden
diff --git a/tests/btrfs/group b/tests/btrfs/group
index a7c6598326c4..7f449d1db99e 100644
--- a/tests/btrfs/group
+++ b/tests/btrfs/group
@@ -233,3 +233,4 @@
228 auto quick volume
229 auto quick send clone
230 auto quick qgroup limit
+231 other
--
2.27.0
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] fstest: random read fio test for read policy
2021-02-22 14:48 [PATCH] fstest: random read fio test for read policy Anand Jain
@ 2021-03-07 17:54 ` Eryu Guan
2021-03-10 23:40 ` Anand Jain
1 sibling, 0 replies; 3+ messages in thread
From: Eryu Guan @ 2021-03-07 17:54 UTC (permalink / raw)
To: Anand Jain; +Cc: fstests, linux-btrfs
On Mon, Feb 22, 2021 at 06:48:18AM -0800, Anand Jain wrote:
> This test case runs fio for raid1/10/1c3/1c4 profiles and all the
> available read policies in the system. At the end of the test case,
> a comparative summary of the result is in the $seqresfull-file.
>
> LOAD_FACTOR parameter controls the fio scalability. For the
> LOAD_FACTOR = 1 (default), this runs fio for file size = 1G and num
> of jobs = 1, which approximately takes 65s to finish.
>
> There are two objectives of this test case. 1. by default, with
> LOAD_FACTOR = 1, it sanity tests the read policies. And 2. Run the
> test case individually with a larger LOAD_FACTOR. For example, 10
> for the comparative study of the read policy performance.
>
> I find tests/btrfs as the placeholder for this test case. As it
> contains many things which are btrfs specific and didn't fit well
> under perf.
>
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
I'd like to let btrfs folks to review as well, to see if this is a sane
test for btrfs.
Thanks,
Eryu
> ---
> tests/btrfs/231 | 145 ++++++++++++++++++++++++++++++++++++++++++++
> tests/btrfs/231.out | 2 +
> tests/btrfs/group | 1 +
> 3 files changed, 148 insertions(+)
> create mode 100755 tests/btrfs/231
> create mode 100644 tests/btrfs/231.out
>
> diff --git a/tests/btrfs/231 b/tests/btrfs/231
> new file mode 100755
> index 000000000000..c08b5826f60a
> --- /dev/null
> +++ b/tests/btrfs/231
> @@ -0,0 +1,145 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2021 Anand Jain. All Rights Reserved.
> +#
> +# FS QA Test 231
> +#
> +# Random read fio test for raid1(10)(c3)(c4) with available
> +# read policy.
> +#
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +fio_config=$tmp.fio
> +fio_results=$tmp.fio_out
> +
> +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 $seqres.full
> +
> +# real QA test starts here
> +
> +# Modify as appropriate.
> +_supported_fs btrfs
> +_require_scratch_dev_pool 4
> +
> +njob=$LOAD_FACTOR
> +size=$LOAD_FACTOR
> +_require_scratch_size $(($size * 2 * 1024 * 1024))
> +echo size=$size njob=$njob >> $seqres.full
> +
> +make_fio_config()
> +{
> + #Set direct IO to true, help to avoid buffered IO so that read happens
> + #from the devices.
> + cat >$fio_config <<EOF
> +[global]
> +bs=64K
> +iodepth=64
> +direct=1
> +invalidate=1
> +allrandrepeat=1
> +ioengine=libaio
> +group_reporting
> +size=${size}G
> +rw=randread
> +EOF
> +}
> +#time_based
> +#runtime=5
> +
> +make_fio_config
> +for job in $(seq 0 $njob); do
> + echo "[foo$job]" >> $fio_config
> + echo "filename=$SCRATCH_MNT/$job/file" >> $fio_config
> +done
> +_require_fio $fio_config
> +cat $fio_config >> $seqres.full
> +
> +work()
> +{
> + raid=$1
> +
> + echo ------------- profile: $raid ---------- >> $seqres.full
> + echo >> $seqres.full
> + _scratch_pool_mkfs $raid >> $seqres.full 2>&1
> + _scratch_mount
> +
> + fsid=$($BTRFS_UTIL_PROG filesystem show -m $SCRATCH_MNT | grep uuid: | \
> + $AWK_PROG '{print $4}')
> + readpolicy_path="/sys/fs/btrfs/$fsid/read_policy"
> + policies=$(cat $readpolicy_path | sed 's/\[//g' | sed 's/\]//g')
> +
> + for policy in $policies; do
> + echo $policy > $readpolicy_path || _fail "Fail to set readpolicy"
> + echo -n "activating readpolicy: " >> $seqres.full
> + cat $readpolicy_path >> $seqres.full
> + echo >> $seqres.full
> +
> + > $fio_results
> + $FIO_PROG --output=$fio_results $fio_config
> + cat $fio_results >> $seqres.full
> + done
> +
> + _scratch_unmount
> + _scratch_dev_pool_put
> +}
> +
> +_scratch_dev_pool_get 2
> +work "-m raid1 -d single"
> +
> +_scratch_dev_pool_get 2
> +work "-m raid1 -d raid1"
> +
> +_scratch_dev_pool_get 4
> +work "-m raid10 -d raid10"
> +
> +_scratch_dev_pool_get 3
> +work "-m raid1c3 -d raid1c3"
> +
> +_scratch_dev_pool_get 4
> +work "-m raid1c4 -d raid1c4"
> +
> +
> +# Now benchmark the raw device performance
> +> $fio_config
> +make_fio_config
> +_scratch_dev_pool_get 4
> +for dev in $SCRATCH_DEV_POOL; do
> + echo "[$dev]" >> $fio_config
> + echo "filename=$dev" >> $fio_config
> +done
> +_require_fio $fio_config
> +cat $fio_config >> $seqres.full
> +
> +echo ------------- profile: raw disk ---------- >> $seqres.full
> +echo >> $seqres.full
> +> $fio_results
> +$FIO_PROG --output=$fio_results $fio_config
> +cat $fio_results >> $seqres.full
> +
> +echo >> $seqres.full
> +echo ===== Summary ====== >> $seqres.full
> +cat $seqres.full | egrep -A1 "Run status|Disk stats|profile:|readpolicy" >> $seqres.full
> +
> +echo "Silence is golden"
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/btrfs/231.out b/tests/btrfs/231.out
> new file mode 100644
> index 000000000000..a31b87a289bf
> --- /dev/null
> +++ b/tests/btrfs/231.out
> @@ -0,0 +1,2 @@
> +QA output created by 231
> +Silence is golden
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index a7c6598326c4..7f449d1db99e 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -233,3 +233,4 @@
> 228 auto quick volume
> 229 auto quick send clone
> 230 auto quick qgroup limit
> +231 other
> --
> 2.27.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] fstest: random read fio test for read policy
2021-02-22 14:48 [PATCH] fstest: random read fio test for read policy Anand Jain
2021-03-07 17:54 ` Eryu Guan
@ 2021-03-10 23:40 ` Anand Jain
1 sibling, 0 replies; 3+ messages in thread
From: Anand Jain @ 2021-03-10 23:40 UTC (permalink / raw)
To: linux-btrfs, dsterba, Josef Bacik; +Cc: fstests
Hi,
How about a review of this test case or suggest any better ideas?
If we are ok with this, I will be adding 2 other types of workloads
that we need to test read policies.
Thanks, Anand
On 22/2/21 10:48 pm, Anand Jain wrote:
> This test case runs fio for raid1/10/1c3/1c4 profiles and all the
> available read policies in the system. At the end of the test case,
> a comparative summary of the result is in the $seqresfull-file.
>
> LOAD_FACTOR parameter controls the fio scalability. For the
> LOAD_FACTOR = 1 (default), this runs fio for file size = 1G and num
> of jobs = 1, which approximately takes 65s to finish.
>
> There are two objectives of this test case. 1. by default, with
> LOAD_FACTOR = 1, it sanity tests the read policies. And 2. Run the
> test case individually with a larger LOAD_FACTOR. For example, 10
> for the comparative study of the read policy performance.
>
> I find tests/btrfs as the placeholder for this test case. As it
> contains many things which are btrfs specific and didn't fit well
> under perf.
>
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
> tests/btrfs/231 | 145 ++++++++++++++++++++++++++++++++++++++++++++
> tests/btrfs/231.out | 2 +
> tests/btrfs/group | 1 +
> 3 files changed, 148 insertions(+)
> create mode 100755 tests/btrfs/231
> create mode 100644 tests/btrfs/231.out
>
> diff --git a/tests/btrfs/231 b/tests/btrfs/231
> new file mode 100755
> index 000000000000..c08b5826f60a
> --- /dev/null
> +++ b/tests/btrfs/231
> @@ -0,0 +1,145 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2021 Anand Jain. All Rights Reserved.
> +#
> +# FS QA Test 231
> +#
> +# Random read fio test for raid1(10)(c3)(c4) with available
> +# read policy.
> +#
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +fio_config=$tmp.fio
> +fio_results=$tmp.fio_out
> +
> +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 $seqres.full
> +
> +# real QA test starts here
> +
> +# Modify as appropriate.
> +_supported_fs btrfs
> +_require_scratch_dev_pool 4
> +
> +njob=$LOAD_FACTOR
> +size=$LOAD_FACTOR
> +_require_scratch_size $(($size * 2 * 1024 * 1024))
> +echo size=$size njob=$njob >> $seqres.full
> +
> +make_fio_config()
> +{
> + #Set direct IO to true, help to avoid buffered IO so that read happens
> + #from the devices.
> + cat >$fio_config <<EOF
> +[global]
> +bs=64K
> +iodepth=64
> +direct=1
> +invalidate=1
> +allrandrepeat=1
> +ioengine=libaio
> +group_reporting
> +size=${size}G
> +rw=randread
> +EOF
> +}
> +#time_based
> +#runtime=5
> +
> +make_fio_config
> +for job in $(seq 0 $njob); do
> + echo "[foo$job]" >> $fio_config
> + echo "filename=$SCRATCH_MNT/$job/file" >> $fio_config
> +done
> +_require_fio $fio_config
> +cat $fio_config >> $seqres.full
> +
> +work()
> +{
> + raid=$1
> +
> + echo ------------- profile: $raid ---------- >> $seqres.full
> + echo >> $seqres.full
> + _scratch_pool_mkfs $raid >> $seqres.full 2>&1
> + _scratch_mount
> +
> + fsid=$($BTRFS_UTIL_PROG filesystem show -m $SCRATCH_MNT | grep uuid: | \
> + $AWK_PROG '{print $4}')
> + readpolicy_path="/sys/fs/btrfs/$fsid/read_policy"
> + policies=$(cat $readpolicy_path | sed 's/\[//g' | sed 's/\]//g')
> +
> + for policy in $policies; do
> + echo $policy > $readpolicy_path || _fail "Fail to set readpolicy"
> + echo -n "activating readpolicy: " >> $seqres.full
> + cat $readpolicy_path >> $seqres.full
> + echo >> $seqres.full
> +
> + > $fio_results
> + $FIO_PROG --output=$fio_results $fio_config
> + cat $fio_results >> $seqres.full
> + done
> +
> + _scratch_unmount
> + _scratch_dev_pool_put
> +}
> +
> +_scratch_dev_pool_get 2
> +work "-m raid1 -d single"
> +
> +_scratch_dev_pool_get 2
> +work "-m raid1 -d raid1"
> +
> +_scratch_dev_pool_get 4
> +work "-m raid10 -d raid10"
> +
> +_scratch_dev_pool_get 3
> +work "-m raid1c3 -d raid1c3"
> +
> +_scratch_dev_pool_get 4
> +work "-m raid1c4 -d raid1c4"
> +
> +
> +# Now benchmark the raw device performance
> +> $fio_config
> +make_fio_config
> +_scratch_dev_pool_get 4
> +for dev in $SCRATCH_DEV_POOL; do
> + echo "[$dev]" >> $fio_config
> + echo "filename=$dev" >> $fio_config
> +done
> +_require_fio $fio_config
> +cat $fio_config >> $seqres.full
> +
> +echo ------------- profile: raw disk ---------- >> $seqres.full
> +echo >> $seqres.full
> +> $fio_results
> +$FIO_PROG --output=$fio_results $fio_config
> +cat $fio_results >> $seqres.full
> +
> +echo >> $seqres.full
> +echo ===== Summary ====== >> $seqres.full
> +cat $seqres.full | egrep -A1 "Run status|Disk stats|profile:|readpolicy" >> $seqres.full
> +
> +echo "Silence is golden"
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/btrfs/231.out b/tests/btrfs/231.out
> new file mode 100644
> index 000000000000..a31b87a289bf
> --- /dev/null
> +++ b/tests/btrfs/231.out
> @@ -0,0 +1,2 @@
> +QA output created by 231
> +Silence is golden
> diff --git a/tests/btrfs/group b/tests/btrfs/group
> index a7c6598326c4..7f449d1db99e 100644
> --- a/tests/btrfs/group
> +++ b/tests/btrfs/group
> @@ -233,3 +233,4 @@
> 228 auto quick volume
> 229 auto quick send clone
> 230 auto quick qgroup limit
> +231 other
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2021-03-10 23:41 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-02-22 14:48 [PATCH] fstest: random read fio test for read policy Anand Jain
2021-03-07 17:54 ` Eryu Guan
2021-03-10 23:40 ` Anand Jain
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.