Stopping Fluentd service on Windows doesn't work well

[tagomoris]

On Windows environment, Fluentd doesn't stop when Administrator operates "stop" for Fluentd Windows Services in Control Panel, which is registered via fluentd --regwinsvc.

When I launched Fluentd process via command prompt, it stops successfully by Ctrl-c.

• Fluentd v0.14.1 (via gem install)
• Windows Server 2012

Configuration file:

<source>
  @type forward
  @label @dump
</source>

<label @dump>
  <match **>
    @type file
    path C:/mydump.*.log
    time_slice_format %Y%m%d%H%M%S
    time_slice_wait 1s
  </match>
</label>

There were no logs when the service was stopped via Services panel, and two "Ruby interpreter" processes were still running.

2016-07-05 07:23:58 +0000 [info]: reading config file path="C:\\fluent_test.conf"
2016-07-05 07:23:58 +0000 [info]: starting fluentd-0.14.1
2016-07-05 07:23:58 +0000 [info]: spawn command to main: C:/Ruby22/bin/ruby.exe
-Eascii-8bit:ascii-8bit  "C:/Ruby22/bin/fluentd" "-c" "C:\fluent_test.conf" "-o" "C:\fluentd.log" "-x" "fluentdwinsvc" --under-supervisor
2016-07-05 07:23:59 +0000 [info]: reading config file path="C:\\fluent_test.conf"
2016-07-05 07:23:59 +0000 [info]: starting fluentd-0.14.1 without supervision
2016-07-05 07:23:59 +0000 [info]: gem 'fluent-plugin-td' version '0.10.28'
2016-07-05 07:23:59 +0000 [info]: gem 'fluentd' version '0.14.1'
2016-07-05 07:23:59 +0000 [info]: gem 'fluentd' version '0.14.0.pre.1'
2016-07-05 07:23:59 +0000 [info]: adding match in @dump pattern="**" type="file"
2016-07-05 07:23:59 +0000 [info]: adding source type="forward"
2016-07-05 07:23:59 +0000 [info]: using configuration file: <ROOT>
  <source>
    @type forward
    @label @dump
  </source>
  <label @dump>
    <match **>
      @type file
      path "C:/mydump.*.log"
      time_slice_format "%Y%m%d%H%M%S"
      time_slice_wait 1s
      buffer_path C:/mydump.*.log
      <buffer time>
        flush_mode lazy
        retry_type exponential_backoff
        path C:/mydump.*.log
        timekey_wait 1s
        timekey 1
      </buffer>
    </match>
  </label>
</ROOT>
2016-07-05 07:23:59 +0000 [info]: listening fluent socket on 0.0.0.0:24224

C:\>

[tagomoris]

It seems that supervisor process uses 25% of CPU, because another process was spawned automatically when I killed a task of 0% CPU usage.

Log about respawning process:

2016-07-05 07:23:59 +0000 [info]: listening fluent socket on 0.0.0.0:24224
2016-07-05 07:29:10 +0000 [info]: Worker 0 finished unexpectedly with status 1
2016-07-05 07:29:11 +0000 [info]: reading config file path="C:\\fluent_test.conf"
2016-07-05 07:29:11 +0000 [info]: starting fluentd-0.14.1 without supervision
2016-07-05 07:29:11 +0000 [info]: gem 'fluent-plugin-td' version '0.10.28'
2016-07-05 07:29:11 +0000 [info]: gem 'fluentd' version '0.14.1'
2016-07-05 07:29:11 +0000 [info]: gem 'fluentd' version '0.14.0.pre.1'
2016-07-05 07:29:11 +0000 [info]: adding match in @dump pattern="**" type="file"
2016-07-05 07:29:11 +0000 [info]: adding source type="forward"
...

[tagomoris]

I found:

• ServerEngine kills workers by KILL signal - it makes shutdown sequence not called (huge problem)
• Even when fluentd is launched on console (non Windows service process), supervisor seems not to kill worker properly