Investigate issue with rdiff-backup restore <defunct>

On Arbuc demo server, I've found the following zombie process. Those should be killed at some point in the process. Let clarify why those process are not killed.

$ ps -ef | grep rdiff
root      2215 29633  0  2021 ?        00:00:00 /usr/bin/python3 /usr/bin/rdiffweb-restore --restore-as-of 1567086158 --encoding utf-8 --kind raw /backups/MyWindowsLaptop/C/Users/vmtest/Desktop/Disks/DOS6.22_bootdisk.iso -
root      2219  2215  0  2021 ?        00:00:00 [rdiff-backup] <defunct>
ikus060  13175 13168  0 16:22 pts/0    00:00:00 grep rdiff
root     29633     1  0  2021 ?        03:37:03 /usr/bin/python3 /usr/bin/rdiffweb
root     30759 29633  0  2021 ?        00:00:00 /usr/bin/python3 /usr/bin/rdiffweb-restore --restore-as-of 1477434528 --encoding utf-8 --kind raw /backups/testcases/이루마 YIRUMA - River Flows in You.mp3 -
root     30763 30759  0  2021 ?        00:00:00 [rdiff-backup] <defunct>

It might be related to the fact the pipe is not properly closed by rdiffweb. Looking at strace of rdiff-backup-restore, the process hang while writing to the pipe.

We need to check if cherrypy is closing the fileobject properly.