A large number of queries often appear 502 bad Gateway

Millions of data were checked directly Nginx 502 bad gateway. Another server is good. I cannot log on to the server to view the configuration. Ask you to get nginx+php comrades. What are the settings of your server? I sent it to the hardware department.

Reply to discussion (solution)

502 is usually the php-fpm of the meal. This may be due to an excessive number of FPM connections or a longer program execution time.
However, you have a sentence to check the millions of data directly Nginx 502 Bad gateway. Lightly, no one can do the remedy.

Insufficient memory, a high number of connections, and a full CPU to handle new connections in a timely manner can cause this 502
That's exactly what you're saying.

There's a way. Server is the hardware over there, I went to ask the situation no one bird I · Now we can only split the data into a small piece of return

Your technical department is too food! Check millions of data at a time.

I this refresh time also appeared 502, also nginx+php,, configuration of the time Nginx inside also has set: Access time (below) and so on several parameters to set the time of these several parameters longer, another may be you in Nginx build time resources may not be enough, hope you have help!
I just happen to have this happen occasionally, let ops restart the service.

Fastcgi_buffer_size 128k;
Fastcgi_buffers 8 128k;
Fastcgi_busy_buffers_size 128k;
Fastcgi_temp_file_write_size 128k;

PHP slow, nginx request half a day not to respond, nature will time out. See if there is a timeout configuration, about fastcgi

Fastcgi_connect_timeout 120;
Fastcgi_send_timeout 120;
Fastcgi_read_timeout 120;
Fastcgi_buffer_size 64k;
Fastcgi_buffers 4 64k;
Fastcgi_busy_buffers_size 128k;
Fastcgi_temp_file_write_size 128k;
Fastcgi_intercept_errors on;
Configuration on the server. I've never changed.

Configuration change (change a bit), the service restart a bit, remember that PHP also has this parameter, you look at

is useless. I think that's the reason for the server setup. Don't change it. Unilateral change useless

This one.... Should be system management to check first .... They found out it was the code problem before it was your turn ...

A server issue has been identified: Hey, no words

Can the landlord describe it in detail? Later confirmed that the server what problem ah, how to find out? How is it solved? Thanks a hell!

