From: Andrei Borzenkov <arvidjaar@gmail.com>
To: "Randy Nürnberger" <ranuberger@posteo.de>,
"Filipe Manana" <fdmanana@kernel.org>
Cc: linux-btrfs@vger.kernel.org
Subject: Re: btrfs send and receive showing errors
Date: Thu, 5 Jan 2023 14:49:38 +0300 [thread overview]
Message-ID: <82ee24b6-fa58-b03e-7765-b157556a2b37@gmail.com> (raw)
In-Reply-To: <58eef5ef-b066-b2cd-e465-6bab43c1c748@posteo.de>
On 05.01.2023 14:33, Randy Nürnberger wrote:
> On Thu, Jan 5, 2023 at 11:42, Filipe Manana wrote:
>> On Thu, Jan 5, 2023 at 10:10 AM Randy Nürnberger <ranuberger@posteo.de> wrote:
>>> On Wed, Jan 4, 2023 at 14:41, Filipe Manana wrote:
>>>> On Wed, Jan 4, 2023 at 1:05 PM Randy Nürnberger <ranuberger@posteo.de> wrote:
>>>>> Hello,
>>>>>
>>>>> I’m in the process of copying a btrfs filesystem (a couple years old)
>>>>> from one disk to another by using btrfs send and receive on all
>>>>> snapshots. The snapshots were created by a tool every hour on the hour
>>>>> as one backup measure and automatically removed as they became older.
>>>>>
>>>>> I got errors like the following and when I compare the copied snapshots
>>>>> with the originals, some files are missing:
>>>>>
>>>>> ERROR: unlink █████ failed: No such file or directory
>>>>> ERROR: link █████ -> █████ failed: No such file or directory
>>>>> ERROR: utimes █████ failed: No such file or directory
>>>>> ERROR: rmdir █████ failed: No such file or directory
>>>>>
>>>>> Is this a known bug and how can I help diagnosing and fixing this?
>>>> So this is a problem caused by the sender issuing commands with outdated paths.
>>>>
>>>> First, try doing the send operation again with a 6.1 kernel - there
>>>> was at least one fix here that may be relevant.
>>> I tried again with the following kernel version and still got the same
>>> errors:
>>> Linux arktos 6.1.0-0-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.2-1~exp1
>>> (2023-01-01) x86_64 GNU/Linux
>>>
>>>> To actual debug things, in case it's not working with 6.1:
>>>>
>>>> 1) Show how you invoked send and receive. I.e. the full command lines
>>>> for 'btrfs send ...' and 'btrfs receive ...'
>>> Those are my full command lines:
>>>
>>> btrfs send -p /mnt/randy/randy-snapshots/2022-01-29--07-00
>>> /mnt/randy/randy-snapshots/2022-02-27--10-00 | btrfs receive -vvv
>>> /mnt/intern/randy-snapshots/ 2>receive-1.txt
>>>
>>> btrfs send -p /mnt/randy/randy-snapshots/2022-02-27--10-00
>>> /mnt/randy/randy-snapshots/2022-03-26--07-00 | btrfs receive -vvv
>>> /mnt/intern/randy-snapshots/ 2>receive-2.txt
>>>
>>>> 2) Provide the whole output of 'btrfs receive' with the -vvv command
>>>> line option.
>>>> This will reveal all path names, but it's necessary to debug things.
>>>> You've hidden path names above, so I suppose that's not acceptable for you.
>>> At least I’m not comfortable sharing the file names on this public
>>> mailing list. I carefully tried to extract and redact what may be the
>>> relevant parts.
>>>
>>> The second command line above is the first one that fails with the
>>> following error: “ERROR: unlink Hase/Fuchs/2022-02-23 Reh.odt failed: No
>>> such file or directory”.
>>>
>>> This is the directory listing for the snapshot before said file was
>>> created (this snapshot can be copied without errors):
>>>
>>> root@arktos /m/r/randy-snapshots# ls -alh 2022-01-29--07-00/Hase/Fuchs/
>>> insgesamt 2,6M
>>> drwxrws--- 1 randy randy 136 19. Dez 2021 ./
>>> drwxrws--- 1 randy randy 134 24. Nov 2021 ../
>>> -rw-rw---- 1 randy randy 38K 5. Mai 2021 '2021-05-01 Igel.odt'
>>> -rw-rw---- 1 randy randy 16K 30. Sep 2021 '2021-09-30 Wolf.odt'
>>> -rw-rw---- 1 randy randy 2,6M 19. Dez 2021 '2021-12-19 Wildschwein.pdf'
>>>
>>> This is the directory listing for the snapshot after the file has been
>>> created (this snapshot can be copied without errors):
>>>
>>> root@arktos /m/r/randy-snapshots# ls -alh 2022-02-27--10-00/Hase/Fuchs/
>>> insgesamt 2,7M
>>> drwxrws--- 1 randy randy 178 27. Feb 2022 ./
>>> drwxrws--- 1 randy randy 134 24. Nov 2021 ../
>>> -rw-rw---- 1 randy randy 38K 5. Mai 2021 '2021-05-01 Igel.odt'
>>> -rw-rw---- 1 randy randy 16K 30. Sep 2021 '2021-09-30 Wolf.odt'
>>> -rw-rw---- 1 randy randy 2,6M 19. Dez 2021 '2021-12-19 Wildschwein.pdf'
>>> -rw-rwx---+ 1 randy randy 42K 26. Feb 2022 '2022-02-23 Reh.odt'*
>>>
>>> This is the directory listing for the snapshot after the file has been
>>> edited in LibreOffice and *renamed* (trying to copy this one fails):
>>>
>>> root@arktos /m/r/randy-snapshots# ls -alh 2022-03-26--07-00/Hase/Fuchs/
>>> insgesamt 2,6M
>>> drwxrws--- 1 randy randy 178 5. Mär 2022 ./
>>> drwxrws--- 1 randy randy 134 24. Nov 2021 ../
>>> -rw-rw---- 1 randy randy 38K 5. Mai 2021 '2021-05-01 Igel.odt'
>>> -rw-rw---- 1 randy randy 16K 30. Sep 2021 '2021-09-30 Wolf.odt'
>>> -rw-rw---- 1 randy randy 2,6M 19. Dez 2021 '2021-12-19 Wildschwein.pdf'
>>> -rw-rw---- 1 randy randy 18K 4. Mär 2022 '2022-03-03 Reh.odt'
>>>
>>> I’ve attached the outputs of the commands above. Apparently ‘btrfs send’
>>> instructs ‘btrfs receive’ to unlink the file ‘Hase/Fuchs/2022-02-23
>>> Reh.odt’ which *does* exist in the copied snapshot ‘2022-02-27--10-00’
>>> and this fails for whatever reason.
Are you sure "btrfs receive" finds the correct parent? Have you ever
flipped read-only property on any snapshot involved in btrfs send/receive?
Check "btrfs subvolume -uqR list" output for duplicated received UUID.
>> The reason is very likely because the file was renamed, but the unlink
>> operation is using the old name (pre-rename name), instead of the new
>> name.
>>
>> For the second receive, there should be other operations affecting the
>> file path Hase/Fuchs/2022-02-23 Reh.odt:
>
> Unfortunately, there are not.
>
> I’m not sure how LibreOffice saves files, but it may create a new file
> and then replace the old one, so the unlink of the old file name can
> make sense, I think.
>
> When I compare the two copied snapshots, the file is indeed gone, so the
> unlink operation actually seems to have worked?
>
> root@arktos /m/i/randy-snapshots# ls -alh 2022-02-27--10-00/Hase/Fuchs/
> insgesamt 2,7M
> drwxrws--- 1 randy randy 178 27. Feb 2022 ./
> drwxrws--- 1 randy randy 134 24. Nov 2021 ../
> -rw-rw---- 1 randy randy 38K 5. Mai 2021 '2021-05-01 Igel.odt'
> -rw-rw---- 1 randy randy 16K 30. Sep 2021 '2021-09-30 Wolf.odt'
> -rw-rw---- 1 randy randy 2,6M 19. Dez 2021 '2021-12-19 Wildschwein.pdf'
> -rw-rwx---+ 1 randy randy 42K 26. Feb 2022 '2022-02-23 Reh.odt'*
>
> root@arktos /m/i/randy-snapshots# ls -alh 2022-03-26--07-00/Hase/Fuchs/
> insgesamt 2,6M
> drwxrws--- 1 randy randy 136 5. Mär 2022 ./
> drwxrws--- 1 randy randy 134 24. Nov 2021 ../
> -rw-rw---- 1 randy randy 38K 5. Mai 2021 '2021-05-01 Igel.odt'
> -rw-rw---- 1 randy randy 16K 30. Sep 2021 '2021-09-30 Wolf.odt'
> -rw-rw---- 1 randy randy 2,6M 19. Dez 2021 '2021-12-19 Wildschwein.pdf'
>
> I also just compiled btrfs-progs v6.1.1 and tried those and still got
> the same error.
>
>>
>> utimes Hase/Fuchs
>> […]
>> unlink Hase/Fuchs/2022-02-23 Reh.odt
>> ERROR: unlink Hase/Fuchs/2022-02-23 Reh.odt failed: No such file or directory
>>
>> Somewhere in the [...] there must be at least one rename of
>> Hase/Fuchs/2022-02-23 Reh.odt into something else.
>> It would be interesting to see more of the receive log to determine if
>> and why that rename happened.
>>
>> If this is blocking you right now, you can do a full send of the
>> snapshot at /mnt/randy/randy-snapshots/2022-03-26--07-00.
>> That will, almost certainly, succeed. Though it will be slower.
>
> Thanks, I’m just copying my filesystem onto bigger disks and that can
> wait some time.
>
>>
>> Thanks.
>>
>>
>>> # sha256sum
>>> /mnt/randy/randy-snapshots/2022-02-27--10-00/Hase/Fuchs/2022-02-23\ Reh.odt
>>> 09ab560f8e2d79e61d253550eda5f388aceddb1b51792e01e589e86a53cdd226
>>>
>>> # sha256sum
>>> /mnt/intern/randy-snapshots/2022-02-27--10-00/Hase/Fuchs/2022-02-23\
>>> Reh.odt
>>> 09ab560f8e2d79e61d253550eda5f388aceddb1b51792e01e589e86a53cdd226
>>>
>>>> Thanks.
>>>>
>>>>> # uname -a # this is the kernel on which this originally happend
>>>>> Linux arktos 5.10.0-20-amd64 #1 SMP Debian 5.10.158-2 (2022-12-13)
>>>>> x86_64 GNU/Linux
>>>>>
>>>>>
>>>>> # uname -a # I already retried everything on the latest Debian
>>>>> backports kernel with the same results
>>>>> Linux arktos 6.0.0-0.deb11.6-amd64 #1 SMP PREEMPT_DYNAMIC Debian
>>>>> 6.0.12-1~bpo11+1 (2022-12-19) x86_64 GNU/Linux
>>>>>
>>>>> # btrfs --version
>>>>> btrfs-progs v5.10.1
>>>>>
>>>>> # btrfs fi sh /mnt/randy # this is the source
>>>>> Label: none uuid: 84bba855-578d-48db-80eb-49f1029c7543
>>>>> Total devices 2 FS bytes used 2.04TiB
>>>>> devid 1 size 4.00TiB used 2.05TiB path /dev/mapper/randy_1_crypt
>>>>> devid 2 size 4.00TiB used 2.05TiB path /dev/mapper/randy_2_crypt
>>>>>
>>>>> # btrfs fi df /mnt/randy
>>>>> Data, RAID1: total=2.02TiB, used=2.02TiB
>>>>> System, RAID1: total=8.00MiB, used=320.00KiB
>>>>> Metadata, RAID1: total=25.00GiB, used=22.99GiB
>>>>> GlobalReserve, single: total=512.00MiB, used=0.00B
>>>>>
>>>>>
>>>>> # btrfs fi sh /mnt/intern # this is the target
>>>>> Label: none uuid: ebb94498-644c-42cd-892f-37886173523c
>>>>> Total devices 2 FS bytes used 1.91TiB
>>>>> devid 1 size 8.00TiB used 1.91TiB path
>>>>> /dev/mapper/vg_arktos_hdd_b-lv_randy_1
>>>>> devid 2 size 8.00TiB used 1.91TiB path
>>>>> /dev/mapper/vg_arktos_hdd_b-lv_randy_2
>>>>>
>>>>> # btrfs fi df /mnt/intern
>>>>> Data, RAID1: total=1.89TiB, used=1.89TiB
>>>>> System, RAID1: total=8.00MiB, used=288.00KiB
>>>>> Metadata, RAID1: total=21.00GiB, used=20.76GiB
>>>>> GlobalReserve, single: total=512.00MiB, used=0.00B
>
next prev parent reply other threads:[~2023-01-05 11:50 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-04 12:54 btrfs send and receive showing errors Randy Nürnberger
2023-01-04 13:41 ` Filipe Manana
2023-01-05 10:10 ` Randy Nürnberger
2023-01-05 10:42 ` Filipe Manana
2023-01-05 11:33 ` Randy Nürnberger
2023-01-05 11:49 ` Andrei Borzenkov [this message]
2023-01-05 15:55 ` Randy Nürnberger
2023-01-05 16:01 ` Andrei Borzenkov
2023-01-05 16:35 ` Randy Nürnberger
2023-01-05 16:47 ` Randy Nürnberger
2023-01-05 17:48 ` Andrei Borzenkov
2023-01-05 18:04 ` Andrei Borzenkov
2023-01-06 11:13 ` Randy Nürnberger
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=82ee24b6-fa58-b03e-7765-b157556a2b37@gmail.com \
--to=arvidjaar@gmail.com \
--cc=fdmanana@kernel.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=ranuberger@posteo.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox