问题
For my symfony project i am using a classic nginx/mysql/php-fpm image combination. For both (fpm and nginx) i use a local mount to the docker container to make the source code available. So that is very slow, but bearable with nfs extension for docker.
In prod environment i prepare the images so, that the local src code is copied to the php-fpm image and a named volume is afterwards created in the docker compose file for the nginx container. So i can connect php-fpm and nginx to make them using the same php files.
Thats basically working.
My problem is, that its still slow to get a site response although i thought that the reason was the slow file mount on my local machine. But still on a native linux production environment, its still slow. And partially slower than on my local dev environment with docker under mac!
For example: - Loading a symfony page displaying 192 Entities on my local machine with mount and nfs mount extension tooks 12 seconds.
On the previouse natively installed prod environment without docker (installed all middleware manually) the same view tooks 1-2 seconds to render.
now on prod environment with docker it takes 16(!) seconds to load this view.
What can i do to run my system faster?
Is the problem still the mount between nginx and php-fpm? Or this this an generally problem with communication over multiple docker container?
One thought was to don't use fpm/nginx combination at all, instead going for apache with php module. So only one container is needed. We are not especially bound to nginx or apache.
Another question: Is it really needful to have the source code in both images (php/nginx)? Or can i save up
Update:
I deinstalled xdebug, that save 1-2 seconds, but its still around 10 seconds ... thats too slow!
For staging its fine, but later for production its very bad ...
来源：https://stackoverflow.com/questions/49009616/linking-nginx-and-php-fpm-container-together-for-fast-interaction-in-docker-prod