Re: btrfs send -p failing: chown o257-1571-0 failed: No such file or directory

From: Filipe Manana <fdmanana@gmail.com>
To: "Massimo B." <massimo.b@gmx.net>
Cc: linux-btrfs <linux-btrfs@vger.kernel.org>
Subject: Re: btrfs send -p failing: chown o257-1571-0 failed: No such file or directory
Date: Wed, 2 Dec 2020 11:04:39 +0000	[thread overview]
Message-ID: <CAL3q7H72N5q6ROhfvuaNNfUvQTe-mtHJVvZaS25oTycJ=3Um3w@mail.gmail.com> (raw)
In-Reply-To: <6ae34776e85912960a253a8327068a892998e685.camel@gmx.net>

On Wed, Dec 2, 2020 at 10:22 AM Massimo B. <massimo.b@gmx.net> wrote:
>
> Hello everybody,
>
> trying to use btrbk archive, it's failing with
> chown o257-1571-0 failed: No such file or directory.
>
> It is doing a btrbk send -p... | btrbk receive ...:
>
> The command is:
> btrfs send -p /mnt/usb/mobiledata/snapshots/mobalindesk/root/root.20200803T060030+0200 /mnt/usb/mobiledata/snapshots/mobalindesk/root/root.20180114T131123+0100 | mbuffer -v 1 -q -m 2% | btrfs receive /mnt/local/data/snapshots/root/
>
> So it takes some newer snapshot as parent to send the older snapshot missing on the target.
> Maybe btrbk just selected the wrong parent?
> Doing the same without -p is working. What's failing here with the parent?
> I tried sending other snapshots with the same parent, but also failing.
> We had several discussions about that in the btrbk tickets with the result, it's
> a btrfs issue.

Yes, it's a btrfs issue.
Pretty much there were always bugs where an incremental send generates
wrong path names depending on how things changed between snapshots,
causing the receiving end to fail.
I have fixed several over the years.

>
> I have 3 btrfs. One is my working one, then I have a local backup and a remote
> backup. Local backup has less snapshots than the remote backup, so I like to
> send some of the remote snapshots back to the local.
>
> Kernel: 5.8.15-gentoo

There were two such bug fixes in 5.9 that are not included in a
vanilla 5.8.15 (dunno if gentoo picked them into their 5.8.15 kernel):

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=98272bb77bf4cc20ed1ffca89832d713e70ebf09
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9c2b4e0347067396ceb3ae929d6888c81d610259

Might be worth trying them, however I think it's probably a different bug.

> Could bees have corrupted something? I'm running it on all 3 btrfs.

Very unlikely, user applications shouldn't be able to do anything that
corrupts a filesystem.

If those don't solve your problem. then the output of 'btrfs receive
-vvv ...' could be used to help debug the issue.

Thanks.

>
> Best regards,
> Massimo
>

-- 
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”