概念
守护进程(Daemon)也称为精灵进程是一种生存期较长的一种进程。它们独立于控制终端并且周期性的执行某种任务或等待处理某些发生的事件。他们常常在系统引导装入时启动,在系统关闭时终止。unix系统有很多守护进程,大多数服务器都是用守护进程实现的,例如inetd守护进程。
需要了解的相关概念
- 进程 (process)
- 进程组 (process group)
- 会话 (session)
可参考以下博文
实现原理
大致流程如下:
- 后台运行
- 脱离控制终端,登录会话和进程组
- 禁止进程重新打开控制终端
- 关闭打开的文件描述符
- 改变当前工作目录
- 重设umask
- 处理SIGCHLD信号
- 日志
关于两次fork
第二个fork不是必须的,只是为了防止进程打开控制终端。
打开一个控制终端的条件是该进程必须是session leader。第一次fork,setsid之后,子进程成为session leader,进程可以打开终端;第二次fork产生的进程,不再是session leader,进程则无法打开终端。
也就是说,只要程序实现得好,控制程序不主动打开终端,无第二次fork亦可。
代码实现
# coding: utf-8
import os
import sys
import time
import atexit
import signal
class Daemon:
def __init__(self, pidfile='/tmp/daemon.pid', stdin='/dev/null', stdout='/dev/null', stderr='/dev/null'):
self.stdin = stdin
self.stdout = stdout
self.stderr = stderr
self.pidfile = pidfile
def daemonize(self):
if os.path.exists(self.pidfile):
raise RuntimeError('Already running.')
# First fork (detaches from parent)
try:
if os.fork() > 0:
raise SystemExit(0)
except OSError as e:
raise RuntimeError('fork #1 faild: {0} ({1})\n'.format(e.errno, e.strerror))
os.chdir('/')
os.setsid()
os.umask(0o22)
# Second fork (relinquish session leadership)
try:
if os.fork() > 0:
raise SystemExit(0)
except OSError as e:
raise RuntimeError('fork #2 faild: {0} ({1})\n'.format(e.errno, e.strerror))
# Flush I/O buffers
sys.stdout.flush()
sys.stderr.flush()
# Replace file descriptors for stdin, stdout, and stderr
with open(self.stdin, 'rb', 0) as f:
os.dup2(f.fileno(), sys.stdin.fileno())
with open(self.stdout, 'ab', 0) as f:
os.dup2(f.fileno(), sys.stdout.fileno())
with open(self.stderr, 'ab', 0) as f:
os.dup2(f.fileno(), sys.stderr.fileno())
# Write the PID file
with open(self.pidfile, 'w') as f:
print(os.getpid(), file=f)
# Arrange to have the PID file removed on exit/signal
atexit.register(lambda: os.remove(self.pidfile))
signal.signal(signal.SIGTERM, self.__sigterm_handler)
# Signal handler for termination (required)
@staticmethod
def __sigterm_handler(signo, frame):
raise SystemExit(1)
def start(self):
try:
self.daemonize()
except RuntimeError as e:
print(e, file=sys.stderr)
raise SystemExit(1)
self.run()
def stop(self):
try:
if os.path.exists(self.pidfile):
with open(self.pidfile) as f:
os.kill(int(f.read()), signal.SIGTERM)
else:
print('Not running.', file=sys.stderr)
raise SystemExit(1)
except OSError as e:
if 'No such process' in str(e) and os.path.exists(self.pidfile):
os.remove(self.pidfile)
def restart(self):
self.stop()
self.start()
def run(self):
pass
使用测试
import os
import sys
import time
from daemon import Daemon
class MyTestDaemon(Daemon):
def run(self):
sys.stdout.write('Daemon started with pid {}\n'.format(os.getpid()))
while True:
sys.stdout.write('Daemon Alive! {}\n'.format(time.ctime()))
sys.stdout.flush()
time.sleep(5)
if __name__ == '__main__':
PIDFILE = '/tmp/daemon-example.pid'
LOG = '/tmp/daemon-example.log'
daemon = MyTestDaemon(pidfile=PIDFILE, stdout=LOG, stderr=LOG)
if len(sys.argv) != 2:
print('Usage: {} [start|stop]'.format(sys.argv[0]), file=sys.stderr)
raise SystemExit(1)
if 'start' == sys.argv[1]:
daemon.start()
elif 'stop' == sys.argv[1]:
daemon.stop()
elif 'restart' == sys.argv[1]:
daemon.restart()
else:
print('Unknown command {!r}'.format(sys.argv[1]), file=sys.stderr)
raise SystemExit(1)
[daemon] python test.py start 23:45:42
[daemon] cat /tmp/daemon-example.pid 23:45:49
8532
[daemon] ps -ef|grep 8532 | grep -v grep 23:46:07
502 8532 1 0 11:45下午 ?? 0:00.00 python test.py start
[daemon] tail -f /tmp/daemon-example.log 23:46:20
Daemon started with pid 8532
Daemon Alive! Fri Dec 2 23:45:49 2016
Daemon Alive! Fri Dec 2 23:45:54 2016
Daemon Alive! Fri Dec 2 23:45:59 2016
Daemon Alive! Fri Dec 2 23:46:04 2016
Daemon Alive! Fri Dec 2 23:46:09 2016
Daemon Alive! Fri Dec 2 23:46:14 2016
Daemon Alive! Fri Dec 2 23:46:19 2016
Daemon Alive! Fri Dec 2 23:46:24 2016
Daemon Alive! Fri Dec 2 23:46:29 2016
Daemon Alive! Fri Dec 2 23:46:34 2016
[daemon] python test.py stop 23:46:36
[daemon] ps -ef|grep 8532 | grep -v grep 23:46:43
参考
- tzuryby/daemon.py python2实现的通用的python daemon类
- 12.14 在Unix系统上面启动守护进程 python3实现的daemon