Nginx configuration resulting in too many connections

Question

In an attempt to implement the upload progress module, the following server configuration is resulting in too many open files error

2014/11/19 12:10:34 [alert] 31761#0: *1010 socket() failed (24: Too many open files) while connecting to upstream, client: 127.0.0.1, server: xxx, request: "GET /documents/15/edit HTTP/1.0", upstream: "http://127.0.0.1:80/documents/15/edit", host: "127.0.0.1"
2014/11/19 12:10:34 [crit] 31761#0: *1010 open() "/usr/share/nginx/html/50x.html" failed (24: Too many open files), client: 127.0.0.1, server: xxx, request: "GET /documents/15/edit HTTP/1.0", upstream: "http://127.0.0.1:80/documents/15/edit", host: "127.0.0.1"

The following is the relevant part of the server bloc which is generating the conflict passenger_enabled on; rails_env development; root /home/user/app/current/public;

    # redirect server error pages to the static page /50x.html
    error_page   500 502 503 504  /50x.html;
    location = /50x.html {
        root   html;
    }

    location / {
        # proxy to upstream server
        proxy_pass http://127.0.0.1;
        proxy_redirect default;

        # track uploads in the 'proxied' zone
        # remember connections for 30s after they finished
        track_uploads proxied 30s;
    }

    location ^~ /progress {
        # report uploads tracked in the 'proxied' zone
        report_uploads proxied;
    }

Being a relative n00b to nginx, I do not comprehend where this is generating the too many files error. I assumed that the error pages are only for 500-504 server errors...

Federate

Answer 1

First check and increase open files limit in the root shell (system-wide, it is 1024 usually):

# ulimit -n
# ulimit -n 16384

Then you can increase nginx's open files limit (main context, top of nginx.conf):

worker_rlimit_nofile 16384

Then restart nginx )

If this not help, check more detail solutions like

Nginx: 24: Too Many Open Files Error And Solution

Schwann

There appears to be some relationship with the worker_connections defined at events level. Because I'm getting 8192 worker_connections are not enough That figure I am pluggin in by trial and error, but resulting in either too many files or insufficient worker_conncetions errors. The link suggests system wide limits. I may try that, but again is there a relationship between hard and soft limits? – Federate