See Stevens and also this lengthy thread on activestate which I found personally to be both mostly incorrect and much to verbose, and I came up with this:
from os import fork, setsid, umask, dup2
from sys import stdin, stdout, stderr
if fork(): exit(0)
umask(0)
setsid()
if fork(): exit(0)
stdout.flush()
stderr.flush()
si = file('/dev/null', 'r')
so = file('/dev/null', 'a+')
se = file('/dev/null', 'a+', 0)
dup2(si.fileno(), stdin.fileno())
dup2(so.fileno(), stdout.fileno())
dup2(se.fileno(), stderr.fileno())
If you need to stop that process again, it is required to know the pid, the usual solution to this is pidfiles. Do this if you need one
from os import getpid
outfile = open(pid_file, 'w')
outfile.write('%i' % getpid())
outfile.close()
For security reasons you might consider any of these after demonizing
from os import setuid, setgid, chdir
from pwd import getpwnam
from grp import getgrnam
setuid(getpwnam('someuser').pw_uid)
setgid(getgrnam('somegroup').gr_gid)
chdir('/')
You could also use nohup but that does not work well with python's subprocess module
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…