Есть две машины которые делят между собой запросы к php. На них крутится nginx и php-fpm.

Обычно один-два раза в логах обоих машин вижу что fpm создает себе в помощь новых детей. Это в принципе нормально только вот при создании пишет:

[26-Feb-2013 17:53:22] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 80 idle, and 208 total children

Вот тут не понятно зачем ему форкатся когда есть 80 ожидающих процессов. Если кто знает поясните пожалуйста.


Второй момент был замечен сегодня утром когда на машинах была минимальная нагрузка. На обоих машинах pfm решил на создавать процессов. На обоих машина в логах примерно такие строчки:

[27-Feb-2013 05:07:58] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 117 idle, and 207 total children
[27-Feb-2013 05:07:59] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 16 children, there are 119 idle, and 210 total children
[27-Feb-2013 05:08:00] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 117 idle, and 211 total children
[27-Feb-2013 05:08:01] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 119 idle, and 214 total children
[27-Feb-2013 05:08:10] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 107 idle, and 223 total children
[27-Feb-2013 05:08:11] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 16 children, there are 109 idle, and 231 total children
[27-Feb-2013 05:08:12] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 117 idle, and 242 total children
[27-Feb-2013 05:08:13] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 110 idle, and 245 total children
[27-Feb-2013 05:08:14] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 118 idle, and 255 total children
[27-Feb-2013 05:08:15] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 117 idle, and 257 total children
[27-Feb-2013 05:08:16] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 116 idle, and 260 total children
[27-Feb-2013 05:08:17] WARNING: [pool default] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 112 idle, and 264 total children


Опять не понятно зачем ему форкатся когда столько ожидающих процессов. Еще не понятно почему он вообще стал форкатся ибо по логам доступа к fpm особой нагрузки не было. Так же график количества подключений в секунду к fpm (и nginx) тоже не показывает аномальной нагрузки. По анализу логов и графиков я убежден что нагрузки не было но тогда почему на обоих машинах в оодно время проявился такой глюк ??