god 初体验 ( my first experience of god)

访问量: 1833

最近的一个项目需要用WEB 来控制 后台运行的进程。 搜索未果,于是在stackoverflow上发了个帖子(http://stackoverflow.com/questions/21069839/whats-the-best-practice-to-control-linux-process-from-web-page-rails)。   同时看到有人推荐godrb.com ,所以试了一下。

据说这个组建可以很好的控制进程的运行,加上各种监控。所以我试了试: 

# /sg552/workspace/test_god/test.rb'
  2 loop do
  3   puts 'hi'
  4   sleep 2
  5 end

# god script
God.watch do |w| 
  w.name = 'hi'
  w.start = 'ruby /sg552/workspace/test_god/test.rb'
  w.keepalive
end

# ouput on ubuntu
sg552@youku:/sg552/workspace/test_god$ god -c hi.god -D
I [2014-01-13 18:12:22]  INFO: Loading hi.god
I [2014-01-13 18:12:22]  INFO: Syslog enabled.
I [2014-01-13 18:12:22]  INFO: Using pid file directory: /home/sg552/.god/pids
I [2014-01-13 18:12:22]  INFO: Socket already in use
I [2014-01-13 18:12:22]  INFO: Socket is stale, reopening
I [2014-01-13 18:12:22]  INFO: Started on drbunix:///tmp/god.17165.sock
I [2014-01-13 18:12:22]  INFO: hi move 'unmonitored' to 'up'
I [2014-01-13 18:12:22]  INFO: hi moved 'unmonitored' to 'up'
I [2014-01-13 18:12:22]  INFO: hi [trigger] process is not running (ProcessRunning)
I [2014-01-13 18:12:22]  INFO: hi move 'up' to 'start'
I [2014-01-13 18:12:22]  INFO: hi start: ruby /sg552/workspace/test_god/test.rb
I [2014-01-13 18:12:22]  INFO: hi moved 'up' to 'up'

god 的NB之处,在于它可以每5秒钟就检测一下对应的进程。 例如,我们看一下目前的进程情况:  

# notice the pid of: 1875,   
sg552@youku:/sg552/workspace/test_god$ ps -ef | grep god
sg552     1702 23782  0 18:20 pts/1    00:00:00 /home/sg552/.rbenv/versions/1.9.3-p327/bin/ruby /home/sg552/.rbenv/versions/1.9.3-p327/bin/god -c hi.god -D
sg552     1875     1  0 18:23 ?        00:00:00 ruby /sg552/workspace/test_god/test.rb
sg552     1885  1763  0 18:24 pts/2    00:00:00 grep --color=auto god

1875 就是god帮我们启动的test.rb 进程, 我们把它干掉试试。。。 

$ kill -9 1875 
# 这个时候, god的输出页面,就会这样, 很神奇吧?:
I [2014-01-13 18:24:13]  INFO: hi [trigger] process is not running (ProcessRunning)
I [2014-01-13 18:24:13]  INFO: hi move 'up' to 'start'
I [2014-01-13 18:24:13]  INFO: hi start: ruby /sg552/workspace/test_god/test.rb
I [2014-01-13 18:24:13]  INFO: hi moved 'up' to 'up'

# 可以看到,它又被启动起来了。
sg552@youku:/sg552/workspace/test_god$ ps -ef | grep test.rb
sg552     1376 23997  0 18:09 pts/4    00:00:00 vim test.rb
sg552     1889     1  0 18:24 ?        00:00:00 ruby /sg552/workspace/test_god/test.rb
sg552     1901  1763  0 18:27 pts/2    00:00:00 grep --color=auto test.rb

看看GOD是如何控制CPU和内存的使用量的:

# 这个脚本比较猛: test.rb
  1 data = ''
  2 loop do
  3   puts 'hi'
  4   100000.times { data << 'x'}
  5 end

# hi.god
God.watch do |w| 
  w.name = 'hi'
  w.start = 'ruby /sg552/workspace/test_god/test.rb'
  w.keepalive(:memory_max => 150.megabytes, :cpu_max => 50.percent )
end

# 重启GOD, 之后,发现CPU越来越高的时候, GOD就会重启这个进程。 
I [2014-01-13 18:48:33]  INFO: hi [trigger] cpu out of bounds [9.345794392462356%%, *84.27876823360998%%, *91.39307897056767%%, *94.0134392181086%%] (CpuUsage)
I [2014-01-13 18:48:33]  INFO: hi move 'up' to 'restart'
I [2014-01-13 18:48:33]  INFO: hi stop: default lambda killer
I [2014-01-13 18:48:33]  INFO: hi sent SIGTERM
I [2014-01-13 18:48:33]  INFO: hi process stopped
I [2014-01-13 18:48:33]  INFO: hi start: ruby /sg552/workspace/test_god/test.rb
I [2014-01-13 18:48:33]  INFO: hi moved 'up' to 'up'

# 干掉这个进程:
$ god stop hi

一些其他的常见问题:

$ god stop hi  # 停止 hi 这个任务(进程)

$ god quit # 彻底停止god

$ god status # 查看god的状态

订阅/RSS Feed

Subscribe

分类/category