From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <amir73il@gmail.com>
MIME-Version: 1.0
In-Reply-To: <20180728084242.33918-3-yi.zhang@huawei.com>
References: <20180728084242.33918-1-yi.zhang@huawei.com> <20180728084242.33918-3-yi.zhang@huawei.com>
From: Amir Goldstein <amir73il@gmail.com>
Date: Mon, 30 Jul 2018 09:24:26 +0300
Message-ID: <CAOQ4uxgqPXZLWMT2gOd9vLDtqA3c6Ma5gfhrFkjONJ6HXHfHZQ@mail.gmail.com>
Subject: Re: [xfstests PATCH 2/3] overlay: add fsck.overlay stress test
Content-Type: text/plain; charset="UTF-8"
To: "zhangyi (F)" <yi.zhang@huawei.com>
Cc: Eryu Guan <guaneryu@gmail.com>, fstests <fstests@vger.kernel.org>, overlayfs <linux-unionfs@vger.kernel.org>, Miklos Szeredi <miklos@szeredi.hu>, Miao Xie <miaoxie@huawei.com>, yangerkun <yangerkun@huawei.com>
List-ID: <linux-unionfs@vger.kernel.org>

On Sat, Jul 28, 2018 at 11:42 AM, zhangyi (F) <yi.zhang@huawei.com> wrote:
> Introduce a test case for fsck.overlay which runs on the underlying
> directories created by fsstress (contain a lot of fs objects) to
> find potential stability issue.
>
> Signed-off-by: zhangyi (F) <yi.zhang@huawei.com>
> ---
>  tests/overlay/061     | 81 +++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/overlay/061.out |  2 ++
>  tests/overlay/group   |  1 +
>  3 files changed, 84 insertions(+)
>  create mode 100755 tests/overlay/061
>  create mode 100644 tests/overlay/061.out
>
> diff --git a/tests/overlay/061 b/tests/overlay/061
> new file mode 100755
> index 0000000..b66b36a
> --- /dev/null
> +++ b/tests/overlay/061
> @@ -0,0 +1,81 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2018 Huawei.  All Rights Reserved.
> +#
> +# FS QA Test No. 061
> +#
> +# Stress test: test fsck.overlay running on the underlying dirs
> +# which were created by fsstress.
> +#
> +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/attr
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +_supported_fs overlay
> +_supported_os Linux
> +_require_scratch_nocheck
> +_require_command "$FSCK_OVERLAY_PROG" fsck.overlay
> +
> +# remove all files from previous tests
> +_scratch_mkfs
> +
> +# Create an underlying layer which contain a lot of random objects
> +create_layer()
> +{
> +       for dir in $*; do
> +               $FSSTRESS_PROG -s 42 -d $dir -z \
> +                       -f creat=20 -f link=10 -f mkdir=20 -f mknod=10 \
> +                       -f rename=10 -f setxattr=10 -f symlink=10 -f write=10 \
> +                       -p 4 -n 500 -l50 > /dev/null 2>&1 &
> +       done
> +}

There is one main difference between this test and overlay/019 -
this test is not sensitive to timing issues and therefore:
1. You gain nothing from running several fsstress processes in parallel
2. You loose something from using a constant ransom seed (-s 42)
current code gives you all layers exactly the same and also
the same for every invocation of test run, which is not what you want.
I think what you want is to use a random seed and record that seed along
with the output of fsstress in $seqres.full so that when test test does catch
a problem with fsck with some random files, you will be able to reproduce
the same layers from the full test output.

Thanks,
Amir.