Thanks for the info, I keep that in mind. For me the content of
"${SRC}" and "${DST}" is of no relevance. They just should point out
what we are dealing with: source or destination side

On Sat, 2021-10-30 at 18:02 +0300, Andrei Borzenkov wrote:

Please understand that mailing list is not a web forum where you can
scroll up and see previous posts. You previous mail may not be
available
when reading so context is lost.

quoted

bash-5.1# btrfs subvolume show "${DST}/@2021-10-30-095620_files"

OK, I interpreted your message wrongly. You were asking for "involved"
snapshots and I assumed those involved in the "buggy" "btrfs
send/receive" command.

I decided to start with new subvolumes (btrfs subvolume create
@files_new; cp -av --reflink @files/ @files_new/; btrfs subvolume
delete @files; mv @files_new @files). Thereafter, I test again.

I asked you to show this information for all snapshots that you have
listed on github for a reason. You so far only provided two source
snapshots and one received snapshot.

Anyway, your two source snapshots show the same received UUID which
confirms what I suspected. It means that every snapshot on
destination
will have the same received UUID as well and "btrfs receive" will use
the wrong snapshot as base to apply change stream.

The same problem was discussed just recently and pops up with
regrettable frequency.

https://lore.kernel.org/linux-btrfs/CAL3q7H5y6z7rRu-ZsZe_WXeHteWx1edZi+L3-UL0aa0oYg+qQA@mail.gmail.com/ (local)