Leave zombie processes once in a while

[onote2]

Environment:

• CentOS release 6.4 (Final) x86_64
• fluentd (0.10.41)
• fluent-plugin-multiprocess (0.1.0)

Log:

# /etc/init.d/td-agent restart
Shutting down td-agent:                                    [  OK  ]
Starting td-agent:                                         [  OK  ]

# ps auxwwf | grep td-agent
root     30277  0.0  0.0 103236   852 pts/0    S+   23:05   0:00                          \_ grep td-agent
td-agent 30190  0.0  0.1 210932 14192 ?        Sl   23:05   0:00 /usr/lib64/fluent/ruby/bin/ruby /usr/sbin/td-agent --group td-agent --log /var/log/td-agent/td-agent.log --daemon /var/run/td-agent/td-agent.pid
td-agent 30193  1.1  0.2 236224 21848 ?        Sl   23:05   0:00  \_ /usr/lib64/fluent/ruby/bin/ruby /usr/sbin/td-agent --group td-agent --log /var/log/td-agent/td-agent.log --daemon /var/run/td-agent/td-agent.pid
td-agent 30202  1.5  0.0      0     0 ?        Z    23:05   0:00      \_ [ruby] <defunct>
td-agent 30215  1.5  0.2 210932 17016 ?        Sl   23:05   0:00      \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24224 \n</source> -c /etc/td-agent/td-agent_child.conf
td-agent 30218  8.3  0.5 323168 41616 ?        Sl   23:05   0:01          \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24224 \n</source> -c /etc/td-agent/td-agent_child.conf

# /etc/init.d/td-agent restart
Shutting down td-agent:                                    [  OK  ]
Starting td-agent:                                         [  OK  ]

# ps auxwwf | grep td-agent
root     30404  0.0  0.0 103236   852 pts/0    S+   23:06   0:00                          \_ grep td-agent
td-agent 30320  0.0  0.1 210932 14188 ?        Sl   23:06   0:00 /usr/lib64/fluent/ruby/bin/ruby /usr/sbin/td-agent --group td-agent --log /var/log/td-agent/td-agent.log --daemon /var/run/td-agent/td-agent.pid
td-agent 30323  1.3  0.2 236224 21784 ?        Sl   23:06   0:00  \_ /usr/lib64/fluent/ruby/bin/ruby /usr/sbin/td-agent --group td-agent --log /var/log/td-agent/td-agent.log --daemon /var/run/td-agent/td-agent.pid
td-agent 30332  1.6  0.2 210932 17016 ?        Sl   23:06   0:00      \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24225 \n</source> -c /etc/td-agent/td-agent_child.conf
td-agent 30338  1.6  0.2 240772 19560 ?        Sl   23:06   0:00      |   \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24225 \n</source> -c /etc/td-agent/td-agent_child.conf
td-agent 30345  1.8  0.2 210932 17012 ?        Sl   23:06   0:00      \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24224 \n</source> -c /etc/td-agent/td-agent_child.conf
td-agent 30348  8.8  0.5 323168 39052 ?        Sl   23:06   0:01          \_ /usr/lib64/fluent/ruby/bin/ruby /usr/lib64/fluent/ruby/lib/ruby/gems/1.9.1/gems/fluentd-0.10.41/lib/fluent/command/fluentd.rb -i <source>\n type forward\n port 24224 \n</source> -c /etc/td-agent/td-agent_child.conf

[Kaliamos]

Hello, Is there some news about this issue ? I am currently encountering it, I use multiprocess to start multiple instances of td-agent for different part of my project. To verify that all my td-agent instances are working well, I check that all my pids are present inside a ps aux, but since even when a subprocess crashes, it leaves a [ruby] <defunct> inside ps aux. So I am not able to know when a subprocess has crashed or not. Is it a normal behaviour ? otherwise do you have any idea about my problem ?

Thank you in advance !