Nginx Tengine 502 Bad Gateway troubleshooting

Gsung should deal with high-concurrency access. I heard that there is a set of methods. To cope with 10 thousand concurrent requests, I got an lnmp for testing. After the installation is complete, the 1000-line concurrent access occurs, and the 502 bad gateway appears. It seems that some configuration files need to be modified accordingly. I found some relevant information on the Internet and shared it with me.

First, the server environment (lnmp) mentioned here refers to linux + nginx + php (php-cgi + php-fpm) + mysql.

502 Bad Gateway is the most frequently encountered problem in lnmp. There are also many causes. The following are the causes of the problem I have collected and their solutions.

1. php process death
The cause is that the error 502 is returned for every dynamic (php) request.
Solution: start php to "service php-fpm start" or "php installation path/php/sbin/php-fpm start"

2. Insufficient number of php-cgi Processes
The cause is that a 502 error occurs in a dynamic request, which is normal. This is because the number of php-cgi processes is insufficient when the concurrency is high.
Solution: Modify the php-fpm configuration php-fpm.conf, increase the number in 60, the max_children number settings not only depend on the server hardware configuration, but also related to the complexity of the php program, generally, the 4G memory can be set to 120-200. Modify the php-fpm reload and modify nginx to configure nginx. conf if necessary. Adjust the number of worker_processes to be equal to the number of CPUs (the default value of worker_processes is 1)

3. php-cgi process timeout
The cause is that the php process encountered a 502 error after running for a period of time. This may be because the php program did not add set_time_limit (0 ), the running time may be limited by the php-fpm settings. The second case is discussed below.
Solution: Modify the php-fpm configuration php-fpm.conf, adjust the number in 30 s to a higher value, adjust to 0 s allows unlimited time to run; if necessary, modify nginx configuration nginx. conf: Increase the parameters fastcgi_connect_timeout, fastcgi_send_timeout, and fastcgi_read_timeout.

4. Insufficient disk space
There are few causes, and the problem is: 502 error or unable to connect to the server. nginx is also dead when the server cannot be connected.
In this case, it takes a long time to start php-fpm or nginx, but the final startup fails. This is probably because the disk space is insufficient. Then you can run the following command to check the disk usage: df-lh. If the disk usage in the root directory "/" is 100%, you need to clear the disk space.
Solution: delete excessive log files and disable logs of some service components. Check the log files in the/var/log directory, nginx and php logs, and mysql logs. If the preceding directories are correct, check other directories one by one (check the/usr directory first) to delete excessive files.

Note: The same applies to Tengine.