Description:
------------
I have hundreds of errors per second appearing in the FPM logs, which suggests that a FPM pool is busy even though PHP was just started seconds ago and there is no traffic yet. This is the second server I have noticed this occurring on, and seems to be recurring continuously since a master commit some time around April. Unfortunately we don't hang on to logs that long.

Below is a snippet of the error message:

########

[18-Jun-2015 10:13:46] NOTICE: fpm is running, pid 24482
[18-Jun-2015 10:13:46] NOTICE: ready to handle connections
[18-Jun-2015 10:13:50] WARNING: [pool nginx] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 0 idle, and 67 total children
[18-Jun-2015 10:13:51] WARNING: [pool nginx] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 16 children, there are 0 idle, and 75 total children
[18-Jun-2015 10:13:52] WARNING: [pool nginx] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 0 idle, and 91 total children
[18-Jun-2015 10:13:53] WARNING: [pool nginx] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 0 idle, and 121 total children
[18-Jun-2015 10:13:54] WARNING: [pool nginx] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 0 idle, and 151 total children
[18-Jun-2015 10:13:55] WARNING: [pool nginx] server reached pm.max_children setting (180), consider raising it
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24542, script '' (request: " ") execution timed out (30.044765 sec), terminating
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24541, script '' (request: " ") execution timed out (30.045377 sec), terminating
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24540, script '' (request: " ") execution timed out (30.045921 sec), terminating
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24539, script '' (request: " ") execution timed out (30.046590 sec), terminating
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24538, script '' (request: " ") execution timed out (30.047270 sec), terminating
.....
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24484 exited on signal 15 (SIGTERM) after 30.090448 seconds from start
[18-Jun-2015 10:14:17] NOTICE: [pool nginx] child 24699 started
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24483 exited on signal 15 (SIGTERM) after 30.094645 seconds from start
[18-Jun-2015 10:14:17] NOTICE: [pool nginx] child 24700 started
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24485 exited on signal 15 (SIGTERM) after 30.094646 seconds from start
[18-Jun-2015 10:14:17] NOTICE: [pool nginx] child 24701 started
[18-Jun-2015 10:14:17] WARNING: [pool nginx] child 24486 exited on signal 15 (SIGTERM) after 30.095189 seconds from start
.....
[ad infinatum]

#######

I would like to point out that nginx is just the name of the default pool in the configuration. That pool is actually never used by the web server to serve requests. Even so, the error is claiming an empty script and request is keeping the pool busy.

It seems as though this issue does not occur with any of the non-default pools that the web server actually uses.

Test script:
---------------
This does not require a PHP script. It occurs 3-4 seconds after FPM start even when no scripts are running.

Expected result:
----------------
No [pool nginx] seems busy messages.
No SIGTERMs.