daemon with restart crash because of _check_sane_and_get_gpg_version

[liz367]

i am using python 3.7.0, gnupg 2.3.1 in centos 7.
i want to build a daemon for my service and add a function that monitors the sigchild signal, and restart my service whenever it crashes

i think the daemon code should be right.
when i add the monitor&restart function, it can work correctly with a simple service:

while True:
    print(1)
    time.sleep(5)

but something wrong when i use gnupg in my service code, even like this simple demo:

time.sleep(3)
print('start')
import gnupg
gnupg.GPG(homedir='/root/.gnupg')
while True:
    print(1)
    time.sleep(5)

the daemon catch so many sigchild signal, and the log show that maybe something go wrong in .../gunpg/_meta/_check_sane_and_get_gpg_version(func)/_read_data(func)/log.debug("Reading data from stream %r..." % stream.repr())

i try to seek for the reason, but i can't handle this question.
but i find if i cancel the function _check_sane_and_get_gpg_version call, the restart function will work.
so i think maybe my daemon code make the proc = self.open_subprocess(["--list-config", "--with-colons"])going wrong.

can anyone help me to figure out the cause?
thanks!
this is my daemon code
daemon.zip