监督(supervisord)下的kafka 0.8.1.1在停止监督守护进程时不会被杀死

Question

在管理员模式下对Kafka的回答后,我试图做同样的事情.事实是,我的kafka(版本0.8.1.1)似乎打包了不同的运行脚本.所以我试图适应并产生以下supervisord配置:

[program:kafka]
command=pidproxy /var/run/kafka.pid /opt/kafka/bin/kafka-server-start.sh /opt/kafka/config/server.properties 
user=kafka
autostart=true
autorestart=true
startsecs=10
startretries=999
log_stdout=true
log_stderr=true
logfile=/var/log/kafka/supervisord-kafka.out
logfile_maxbytes=20MB
logfile_backups=10

我使用以下rpm的supervisor:

[root@dev-xxx yyy]# rpm -qi supervisor
Name        : supervisor                   Relocations: (not relocatable)
Version     : 3.0                               Vendor: obs://build.opensuse.org/home:presbrey
Release     : 13.1                          Build Date: Thu 13 Jun 2013 08:35:40 AM IDT

停止监督守护进程并没有杀死卡夫卡,见下文:

[root@dev-xxx yyy]# jps
3236 core
31378 QuorumPeerMain
31408 Jps
2964 Kafka
31293 supervisor
31246 nimbus

[root@dev-xxx yyy]# supervisorctl status
kafka                            RUNNING    pid 2963, uptime 1 day, 5:15:18
storm-nimbus                     RUNNING    pid 31246, uptime 0:02:06
storm-supervisor                 RUNNING    pid 31293, uptime 0:01:44
storm-ui                         RUNNING    pid 3236, uptime 5 days, 22:37:33
storm-zookeeper                  RUNNING    pid 31378, uptime 0:00:47


[root@dev-xxx yyy]# service supervisord stop
Stopping supervisord: Shut down
Waiting roughly 60 seconds for /var/run/supervisord.pid to be removed after child processes exit
Supervisord exited as expected in under 12 seconds


[root@dev-xxx yyy]# supervisorctl status
unix:///var/tmp/supervisor.sock no such file


[root@dev-xxx yyy]# jps
2964 Kafka
31474 Jps

Answer 1

好吧,正如我之前所说,stopasgroup=true应该有所帮助.我不是100%肯定确切的原因,但是根据我的理解,这是因为kafka是由shell脚本启动的,默认情况下,在拆除时,只有该脚本会被杀死.