In general I have no ideas, but I do know that scripts executed
in udev rules have to run "fast", or they are automatically
killed off. You might want to consider having the script
do something like:
/usr/bin/at -M now <<'HERE' > /dev/null 2>&1
/full/path/to/do-the-backup/script
HERE
That will very quickly arrange for a different thread
of control to run the actual backup, and avoid systemd
killing it off because it is too slow.