differential snapshot send fails with: cannot find parent subvolume c064acd9-44…

[jschwender]

Error when sending diff snapshot to external disk:

[root@jupiter]# btrfs send --proto 2 -p  /home/joachim/Test/.snapshots/6/snapshot  /home/joachim/Test/.snapshots/7/snapshot/ | btrfs receive /mnt/backup-bt.extern/backup/root/
At subvol /home/joachim/Test/.snapshots/7/snapshot/
At snapshot snapshot
ERROR: snapshot: cannot find parent subvolume c064acd9-44…

This is in contradiction with documentation, therefore fulfills the definition of a defect.

same diff snapshot sent to the same disk works like expected:

[root@jupiter]# btrfs send --proto 2 -p  /home/joachim/Test/.snapshots/6/snapshot  /home/joachim/Test/.snapshots/7/snapshot/ | btrfs receive /home
At subvol /home/joachim/Test/.snapshots/7/snapshot/
At snapshot snapshot

btrfs --version
btrfs-progs v6.15
-EXPERIMENTAL -INJECT -STATIC +LZO +ZSTD +UDEV +FSVERITY +ZONED CRYPTO=libgcrypt

uname -a
Linux jupiter 6.15.4-arch2-1 #1 SMP PREEMPT_DYNAMIC Fri, 27 Jun 2025 16:35:07 +0000 x86_64 GNU/Linux

[qxsnap]

The parent subvolume you've specified doesn't exist in the destination filesystem. Incremental mode requires parent snapshots to already be present at the destination. Do a full transfer of the parent snapshot, then incremental send will work.

From btrfs-send(8):
In the incremental mode (options -p and -c), previously sent snapshots that are available on both the sending and receiving side can be used to reduce the amount of information that has to be sent to reconstruct the sent snapshot on a different filesystem.

It works to the same disk because the parent snapshot is already on that disk.

[jschwender]

Thank you, got it. Very helpful. So the error message seems to be emitted by receive? In my opinion, the error message is exclusively for expert level ++. Not much compatible with a end user perspective. It should state clearly what and where it does not find something. It may be systematically correct (for the expert++) to refer to a volume by UUID, but the user does not specify by UUID, he gives a path in the command. Consequently, it would be nice if the error message would refer to the path the user specified, and also that the error was created by the receiver part of the command (if my guess is correct) and make clear that it is missing the volume on the target/receiver side.

[qxsnap]

Correct, the error is from btrfs receive, which doesn't have any additional metadata about the parent snapshot, aside from the UUID.

Take a look at the command stream using

btrfs send --no-data --proto 2 -p  /home/joachim/Test/.snapshots/6/snapshot  /home/joachim/Test/.snapshots/7/snapshot/ | btrfs receive --dump | head

You should see the first command from btrfs send is "snapshot" which includes the parent_uuid that btrfs receive will clone on the destination filesystem as the base from which to apply the subsequent commands describing the incremental changes:

btrfs receive fails at this first command, because it can't locate a subvolume with UUID=parent_uuid on the receiving filesystem. Agree this would be clearer if the error explicitly showed as ERROR receive:.

[jschwender]

ok. thank you for your patience. see pull request #1005