Re: [PATCH 2/2] btrfs: Add test for btrfs balance convert functionality

[Nikolay Borisov <nborisov@suse.com>]

On 27.09.19 г. 15:22 ч., Qu Wenruo wrote:
> 
> 
> On 2019/9/27 下午7:50, Nikolay Borisov wrote:
>>
>>
>> On 27.09.19 г. 14:21 ч., Qu Wenruo wrote:
>>>
>>>
>>> On 2019/9/27 下午6:52, Nikolay Borisov wrote:
>>>> This does an exhaustive testing of all possible conversion combination.
>>>>
>>>> Signed-off-by: Nikolay Borisov <nborisov@suse.com>
>>>> ---
>>>>
>>>> This is a rather long test - it takes around 38 minutes, OTOH it exercies around
>>>> 1780 combinations of source/destination test.
>>>
>>> Exactly the problem I'm concerning.
>>>
>>> However we all know that btrfs convert works by relocating old data to
>>> new chunks.
>>> It means the source doesn't matter that much.
>>>
>>> As long as the chunk read code works fine, converting from single to
>>> RAID10 is not that different from converting from DUP to RAID10.
>>> (ALthough there is still some difference due to different nr_disks and
>>> dev extent layouts, but that's not the core problem)
>>>
>>> By that we can change from testing all the combinations to just testing
>>> all destination profiles.
>>>
>>> This should only needs about 6 tests, and you can reuse all the same
>>> setup to fulfill all tests.
>>
>> True, but thanks to the exhaustive tests I was able to catch xfstest
>> special casing -mdup as source argument which resulted in patch 1 of
>> this series. I will leave that here to gather some more feedback and
>> will trim down the tests.
>>
>> And regarding the number of tests - do we want to mix the source
>> profiles of data/metadata.
> 
> To me, unless we have some strong evident in how different data/metadata
> profiles can cause different behavior, then using the same profile
> should be OK.
> 
>> Because it's true that it takes 6 test to
>> convert from
>>
>> SINGLE=>DUP, RAID1, RAID5, RAID0, RAID10, RAID6
>> but we also need a 7th test e.g. DUP->SINGLE.
> 
> Ah, I forgot RAID6. Then it's indeed 7 tests.
> 
> BTW, with 7 tests, we can afford more extensive tests, like 15~30s
> fsstress at the background, and after balance run a full scrub, then
> umount and fsck.

Makes sense I will work in that direction.

> 
> Thanks,
> Qu
> 
>>
>>>
>>> Just 4 devices, then you can go convert to SINGLE, DUP, RAID1, RAID5,
>>> RAID6, RAID10.
>>>
>>> Thanks,
>>> Qu
>>>
>>>>
>>>>
>>>>  tests/btrfs/194     | 1843 +++++++++++++++++++++++++++++++++++++++++++++++++++
>>>>  tests/btrfs/194.out |    2 +
>>>>  tests/btrfs/group   |    1 +
>>>>  3 files changed, 1846 insertions(+)
>>>>  create mode 100755 tests/btrfs/194
>>>>  create mode 100644 tests/btrfs/194.out
>>>>
>>>> diff --git a/tests/btrfs/194 b/tests/btrfs/194
>>>> new file mode 100755
>>>> index 000000000000..7ba4555c12b0
>>>> --- /dev/null
>>>> +++ b/tests/btrfs/194
>>>> @@ -0,0 +1,1843 @@
>>>> +#! /bin/bash
>>>> +# SPDX-License-Identifier: GPL-2.0
>>>> +# Copyright (c) 2019 SUSE Linux Products GmbH. All Rights Reserved.
>>>> +#
>>>> +# FS QA Test 194
>>>> +#
>>>> +# Exercises all available combinations of btrfs balance start -d/-m convert
>>>> +#
> [...]
>>>> +
>>>> +for i in "${TEST_VECTORS[@]}"; do
>>>> +	run_testcase $i
>>>> +done
>>>> +
>>>> +echo "Silence is golden"
>>>> +status=0
>>>> +exit
>>>> diff --git a/tests/btrfs/194.out b/tests/btrfs/194.out
>>>> new file mode 100644
>>>> index 000000000000..7bfd50ffb5a4
>>>> --- /dev/null
>>>> +++ b/tests/btrfs/194.out
>>>> @@ -0,0 +1,2 @@
>>>> +QA output created by 194
>>>> +Silence is golden
>>>> diff --git a/tests/btrfs/group b/tests/btrfs/group
>>>> index b92cb12ca66f..a2c0ad87d0f6 100644
>>>> --- a/tests/btrfs/group
>>>> +++ b/tests/btrfs/group
>>>> @@ -196,3 +196,4 @@
>>>>  191 auto quick send dedupe
>>>>  192 auto replay snapshot stress
>>>>  193 auto quick qgroup enospc limit
>>>> +194 auto volume balance
>>>>
>>>
>