nginx - php-fpm crashing -
I run an API using RakeSpace cloud server. For our peak time, we have an auto-level policy and we think that one day or twice a day that the servers roam, they are having problems with php-fpm.
As soon as the server starts receiving traffic, all 40 children of it are born and then 502 starts with error message and NGNEX starts. I can monitor the children at the top and they sit idle for about 5 minutes and then eventually die and the server can get traffic again, but then the same process starts and it locks again. .
We are running another server that has the exact same disk image, same configuration as it is and it will behave completely. Below is my php5-fpm config. Any help would be greatly appreciated.
[www.conf] Listen = 127.0.0.1:9000 pm = Speed pm.max_children = 40 pm.start_servers = 10 pm.min_spare_servers = 10 pm.max_spare_servers = 10 pm.max_requests = 200 php-fpm Check 'listen.backlog' for each pool configuration option (default) : -1 = unlimited). Since sockets are being used, this value is limited by the system value 'net.core.somaxconn' (default: 128 on most Linux systems).
Comments
Post a Comment