Difference between revisions of "Track HTTP Application Response time in Nginx"

From wikieduonline
Jump to navigation Jump to search
Tags: Mobile web edit, Mobile edit
 
(22 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
<ref>https://lincolnloop.com/blog/tracking-application-response-time-nginx/</ref>
 
<ref>https://lincolnloop.com/blog/tracking-application-response-time-nginx/</ref>
  
1) Add to <code>/etc/nginx/[[nginx.conf]]</code>
+
1) Add to <code>/etc/nginx/[[nginx.conf]]</code> befor your <code>[[access_log]]</code> directive http://nginx.org/en/docs/http/ngx_http_log_module.html
 
 
http://nginx.org/en/docs/http/ngx_http_log_module.html
 
  
 +
<pre>
 
  log_format timed_combined '$remote_addr - $remote_user [$time_local] '
 
  log_format timed_combined '$remote_addr - $remote_user [$time_local] '
 
     '"$request" $status $body_bytes_sent '
 
     '"$request" $status $body_bytes_sent '
 
     '"$http_referer" "$http_user_agent" '
 
     '"$http_referer" "$http_user_agent" '
     '$request_time $upstream_response_time $pipe';
+
     '$pipe $upstream_response_time $request_time';
   
+
  </pre>
  
2) Modify access_log directive to use new format:
+
2) Modify <code>[[access_log]]</code> directive to use new format:
 
<pre>
 
<pre>
 
access_log /var/log/nginx/yourdomain.com.access.log timed_combined;
 
access_log /var/log/nginx/yourdomain.com.access.log timed_combined;
Line 20: Line 19:
 
:<code>pipe</code> Shows ‘p’ in case the request was pipelined.
 
:<code>pipe</code> Shows ‘p’ in case the request was pipelined.
  
 +
 +
3) Reload nginx:
 +
[[nginx -t]] && [[systemctl restart]] nginx
 +
 +
4) Check your new log format, last field contains (<code>$request_time</code>):
 +
[[tail]] -f /var/log/nginx/[[access.log]]
 +
 +
== Related ==
 +
* Print http response times and date:
 +
:<code>cat [[/var/log/nginx/access.log]] | awk '{print $4, $([[NF-1]])}'</code>
 +
 +
Print only if http response time exceed a threshold:
 +
:<code>cat /var/log/nginx/access.log | awk '$(NF-1) > "0.300" {print $4, $(NF-1)}</code>
 +
 +
Web site [[response time]]:
 +
* <code>[[curl -so]] /dev/null -w '%{time_total}\n' https://google.com</code>
  
 
== See also ==
 
== See also ==
 
* {{Nginx}}
 
* {{Nginx}}
 +
 +
 +
[[Category:Nginx]]

Latest revision as of 07:47, 9 May 2023

[1]

1) Add to /etc/nginx/nginx.conf befor your access_log directive http://nginx.org/en/docs/http/ngx_http_log_module.html

 log_format timed_combined '$remote_addr - $remote_user [$time_local] '
    '"$request" $status $body_bytes_sent '
    '"$http_referer" "$http_user_agent" '
    '$pipe $upstream_response_time $request_time';
 

2) Modify access_log directive to use new format:

access_log /var/log/nginx/yourdomain.com.access.log timed_combined;

[2]

request_time This shows how long Nginx dealt with the request
upstream_response_time Gives us the time it took our upstream server (in this case Apache/mod_wsgi) to respond
pipe Shows ‘p’ in case the request was pipelined.


3) Reload nginx:

nginx -t && systemctl restart nginx

4) Check your new log format, last field contains ($request_time):

tail -f /var/log/nginx/access.log

Related[edit]

  • Print http response times and date:
cat /var/log/nginx/access.log | awk '{print $4, $(NF-1)}'

Print only if http response time exceed a threshold:

cat /var/log/nginx/access.log | awk '$(NF-1) > "0.300" {print $4, $(NF-1)}

Web site response time:

See also[edit]

  • https://lincolnloop.com/blog/tracking-application-response-time-nginx/
  • https://stackoverflow.com/a/39260524
  • Advertising: