转自:http://blog.csdn.net/zyz511919766/article/details/43967793
supervisor:C/S架构的进程控制系统,可使用户在类UNIX系统中监控、管理进程。常用于管理与某个用户或项目相关的进程。
组成部分
supervisord:服务守护进程
supervisorctl:命令行客户端
Web Server:提供与supervisorctl功能相当的WEB操作界面
XML-RPC Interface:XML-RPC接口
安装
centos平台下可直接用过YUM源安装
yum info supervisor
sudo yum install supervisor
sudo chkconfig supervisord on
服务器启停
sudo /etc/init.d/supervisord {start|stop|status|restart|reload|force-reload|condrestart}
日志
/var/log/supervisor/supervisord.log
配置文件
sudo vim /etc/supervisord.conf
需要重点关注的是以部分
[program:x]中配置要监控的进程
配置样例
- 
[supervisord]  
 - 
http_port=/var/tmp/supervisor.sock ; (default is to run a UNIX domain socket server)  
 - 
;http_port=127.0.0.1:9001  ; (alternately, ip_address:port specifies AF_INET)  
 - 
;sockchmod=0700              ; AF_UNIX socketmode (AF_INET ignore, default 0700)  
 - 
;sockchown=nobody.nogroup     ; AF_UNIX socket uid.gid owner (AF_INET ignores)  
 - 
;umask=022                   ; (process file creation umask;default 022)  
 - 
logfile=/var/log/supervisor/supervisord.log ; (main log file;default $CWD/supervisord.log)  
 - 
logfile_maxbytes=50MB       ; (max main logfile bytes b4 rotation;default 50MB)  
 - 
logfile_backups=10          ; (num of main logfile rotation backups;default 10)  
 - 
loglevel=info               ; (logging level;default info; others: debug,warn)  
 - 
pidfile=/var/run/supervisord.pid ; (supervisord pidfile;default supervisord.pid)  
 - 
nodaemon=false              ; (start in foreground if true;default false)  
 - 
minfds=1024                 ; (min. avail startup file descriptors;default 1024)  
 - 
minprocs=200                ; (min. avail process descriptors;default 200)  
 - 
  
 - 
;nocleanup=true              ; (don‘t clean up tempfiles at start;default false)  
 - 
;http_username=user          ; (default is no username (open system))  
 - 
;http_password=123           ; (default is no password (open system))  
 - 
;childlogdir=/tmp            ; (‘AUTO‘ child log dir, default $TEMP)  
 - 
;user=chrism                 ; (default is current user, required if root)  
 - 
;directory=/tmp              ; (default is not to cd during start)  
 - 
;environment=KEY=value       ; (key value pairs to add to environment)  
 - 
  
 - 
[supervisorctl]  
 - 
serverurl=unix:///var/tmp/supervisor.sock ; use a unix:// URL  for a unix socket  
 - 
;serverurl=http://127.0.0.1:9001 ; use an http:// url to specify an inet socket  
 - 
;username=chris              ; should be same as http_username if set  
 - 
;password=123                ; should be same as http_password if set  
 - 
;prompt=mysupervisor         ; cmd line prompt (default "supervisor")  
 - 
  
 - 
; The below sample program section shows all possible program subsection values,  
 - 
; create one or more ‘real‘ program: sections to be able to control them under  
 - 
; supervisor.  
 - 
  
 - 
;[program:example]  
 - 
;command=/bin/echo; the program (relative uses PATH, can take args)  
 - 
;priority=999                ; the relative start priority (default 999)  
 - 
;autostart=true              ; start at supervisord start (default: true)  
 - 
;autorestart=true            ; retstart at unexpected quit (default: true)  
 - 
;startsecs=10                ; number of secs prog must stay running (def. 10)  
 - 
;startretries=3              ; max # of serial start failures (default 3)  
 - 
;exitcodes=0,2               ; ‘expected‘ exit codes for process (default 0,2)  
 - 
;stopsignal=QUIT             ; signal used to kill process (default TERM)  
 - 
;stopwaitsecs=10             ; max num secs to wait before SIGKILL (default 10)  
 - 
;user=chrism                 ; setuid to this UNIX account to run the program  
 - 
;log_stdout=true             ; if true, log program stdout (default true)  
 - 
;log_stderr=true             ; if true, log program stderr (def false)  
 - 
;logfile=/var/log/supervisor.log    ; child log path, use NONE for none; default AUTO  
 - 
;logfile_maxbytes=1MB        ; max # logfile bytes b4 rotation (default 50MB)  
 - 
;logfile_backups=10          ; # of logfile backups (default 10)  
 
 
“;”为注释。各参数的含义都很明确。可以根据官方手册结合实验来进一步深入了解。重点说几个[program:example]中的参数
- 
;command=/bin/echo;         supervisor启动时将要开启的进程。相对或绝对路径均可。若是相对路径则会从supervisord的$PATH变中查找。命令可带参数。  
 - 
;priority=999                   指明进程启动和关闭的顺序。低优先级表明进程启动时较先启动关闭时较后关闭。高优先级表明进程启动时启动时较后启动关闭时较先关闭。  
 - 
;autostart=true                 是否随supervisord启动而启动  
 - 
;autorestart=true               进程意外退出后是否自动重启  
 - 
;startsecs=10                   进程持续运行多久才认为是启动成功  
 - 
;startretries=3                 重启失败的连续重试次数  
 - 
;exitcodes=0,2                  若autostart设置为unexpected且监控的进程并非因为supervisord停止而退出,那么如果进程的退出码不在exitcode列表中supervisord将重启进程  
 - 
;stopsignal=QUIT                杀进程的信号  
 - 
;stopwaitsecs=10                向进程发出stopsignal后等待OS向supervisord返回SIGCHILD 的时间。若超时则supervisord将使用SIGKILL杀进程  
 
 
一个Rabbitmq项目中生产者和消费者进程使用supervisor监控的配置情况:(配置中的其他部分略)
- 
</pre><p><pre name="code" class="plain">[program:worker_for_summary]  
 - 
command=/home/op1/scripts/rabbitmqclient/worker_for_summary.py  
 - 
priority=1  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:worker_for_detail_all]  
 - 
command=/home/op1/scripts/rabbitmqclient/worker_for_detail_all.py  
 - 
priority=1  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:worker_for_detail_recent_list]  
 - 
command=/home/op1/scripts/rabbitmqclient/worker_for_detail_recent_list.py  
 - 
priority=1  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:worker_for_detail_recent_sset]  
 - 
command=/home/op1/scripts/rabbitmqclient/worker_for_detail_recent_sset.py  
 - 
priority=1  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:publisher_for_summary]  
 - 
command=/home/op1/scripts/rabbitmqclient/publisher_for_summary.py  
 - 
priority=999  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:publisher_for_summary_nt]  
 - 
command=/home/op1/scripts/rabbitmqclient/publisher_for_summary_nt.py  
 - 
priority=999  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:publisher_for_detail]  
 - 
command=/home/op1/scripts/rabbitmqclient/publisher_for_detail.py  
 - 
priority=999  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
  
 - 
[program:publisher_for_detail_nt]  
 - 
command=/home/op1/scripts/rabbitmqclient/publisher_for_detail_nt.py  
 - 
priority=999  
 - 
log_stderr=true             ; if true, log program stderr (def false)  
 - 
                                                                      
 
 
配置完成后启动supervisord
- 
sudo /etc/init.d/supervisord start  
 
 
可以看到配置的各个进程在后台运行了起来。
停掉某个进程后supervisor会马上重启该进程
停止supervisor
- 
sudo /etc/init.d/supervisord stop  
 
 
可以看到配置的各个进程都停止运行了。
可以通过supervisorctl查看管理监控的进程情况:
- 
[op1@SVR1631HP360 ~]$ sudo supervisorctl  
 - 
publisher_for_detail RUNNING    pid 27557, uptime 0:00:45  
 - 
publisher_for_detail_nt RUNNING    pid 27567, uptime 0:00:45  
 - 
publisher_for_summary RUNNING    pid 27566, uptime 0:00:45  
 - 
publisher_for_summary_nt RUNNING    pid 27568, uptime 0:00:45  
 - 
worker_for_detail_all RUNNING    pid 27581, uptime 0:00:45  
 - 
worker_for_detail_recent RUNNING    pid 27582, uptime 0:00:45  
 - 
worker_for_summary RUNNING    pid 27559, uptime 0:00:45  
 - 
  
 - 
#可通过help了解命令的更多用法  
 - 
supervisor> help  
 - 
  
 - 
Documented commands (type help <topic>):  
 - 
========================================  
 - 
EOF    exit  maintail  quit    restart   start   stop  
 - 
clear  help  open      reload  shutdown  status  tail  
 - 
  
 - 
supervisor> help stop  
 - 
stop <processname>            Stop a process.  
 - 
stop <processname> <processname>    Stop multiple processes  
 - 
stop all                Stop all processes  
 - 
  When all processes are stopped, they are stopped in  
 - 
  reverse priority order (see config file)  
 - 
supervisor> help status  
 - 
status          Get all process status info.  
 - 
status <name>     Get status on a single process by name.  
 - 
status <name> <name>    Get status on multiple named processes.  
 - 
  
 - 
#停止某个进程  
 - 
supervisor> stop publisher_for_summary  
 - 
publisher_for_summary: stopped  
 - 
  
 - 
#查看此时此刻的状态  
 - 
supervisor> status  
 - 
publisher_for_detail RUNNING    pid 27557, uptime 0:05:41  
 - 
publisher_for_detail_nt RUNNING    pid 27567, uptime 0:05:41  
 - 
publisher_for_summary STOPPED    Feb 27 02:48 PM  
 - 
publisher_for_summary_nt RUNNING    pid 27568, uptime 0:05:41  
 - 
worker_for_detail_all RUNNING    pid 27581, uptime 0:05:41  
 - 
worker_for_detail_recent RUNNING    pid 27582, uptime 0:05:41  
 - 
worker_for_summary RUNNING    pid 27559, uptime 0:05:41  
 - 
#发现被supervisorctl停掉的进程不会被自动重启  
 - 
  
 - 
#开启刚才停掉的进程  
 - 
supervisor> start publisher_for_summary  
 - 
publisher_for_summary: started  
 - 
supervisor> status  
 - 
publisher_for_detail RUNNING    pid 27557, uptime 0:08:02  
 - 
publisher_for_detail_nt RUNNING    pid 27567, uptime 0:08:02  
 - 
publisher_for_summary RUNNING    pid 3035, uptime 0:00:04  
 - 
publisher_for_summary_nt RUNNING    pid 27568, uptime 0:08:02  
 - 
worker_for_detail_all RUNNING    pid 27581, uptime 0:08:02  
 - 
worker_for_detail_recent RUNNING    pid 27582, uptime 0:08:02  
 - 
worker_for_summary RUNNING    pid 27559, uptime 0:08:02  
 - 
  
 - 
#停掉所有进程  
 - 
supervisor> stop all  
 - 
worker_for_detail_recent: stopped  
 - 
worker_for_detail_all: stopped  
 - 
publisher_for_summary_nt: stopped  
 - 
publisher_for_detail_nt: stopped  
 - 
publisher_for_summary: stopped  
 - 
worker_for_summary: stopped  
 - 
publisher_for_detail: stopped  
 - 
supervisor> status  
 - 
publisher_for_detail STOPPED    Feb 27 02:51 PM  
 - 
publisher_for_detail_nt STOPPED    Feb 27 02:51 PM  
 - 
publisher_for_summary STOPPED    Feb 27 02:51 PM  
 - 
publisher_for_summary_nt STOPPED    Feb 27 02:51 PM  
 - 
worker_for_detail_all STOPPED    Feb 27 02:51 PM  
 - 
worker_for_detail_recent STOPPED    Feb 27 02:51 PM  
 - 
worker_for_summary STOPPED    Feb 27 02:51 PM  
 - 
  
 - 
#开启所有进程  
 - 
supervisor> start all  
 - 
publisher_for_detail: started  
 - 
worker_for_summary: started  
 - 
publisher_for_summary: started  
 - 
publisher_for_detail_nt: started  
 - 
publisher_for_summary_nt: started  
 - 
worker_for_detail_all: started  
 - 
worker_for_detail_recent: started  
 - 
supervisor> status  
 - 
publisher_for_detail RUNNING    pid 5111, uptime 0:00:15  
 - 
publisher_for_detail_nt RUNNING    pid 5141, uptime 0:00:15  
 - 
publisher_for_summary RUNNING    pid 5135, uptime 0:00:15  
 - 
publisher_for_summary_nt RUNNING    pid 5147, uptime 0:00:15  
 - 
worker_for_detail_all RUNNING    pid 5153, uptime 0:00:15  
 - 
worker_for_detail_recent RUNNING    pid 5159, uptime 0:00:14  
 - 
worker_for_summary RUNNING    pid 5112, uptime 0:00:15  
 
 
更多内容请参考官方手册
http://supervisord.org/
supervisor安装配置与使用
原文:http://blog.csdn.net/bluehawksky/article/details/50978755