Servers statistics

Server statistics

A file with statistics per each backend server is created in /proc/tempesta/servers/ directory. This file can be seen as follows:

# cat /proc/tempesta/servers/192.168.10.230\:8080 
Minimal response time           : 0ms
Average response time           : 4ms
Median  response time           : 3ms
Maximum response time           : 66ms
Percentiles
50%:    3ms
75%:    7ms
90%:    11ms
95%:    15ms
99%:    29ms
Total pinned sessions           : 0
Total schedulable connections   : 3
HTTP health monitor is enabled  : 1
HTTP availability       : 1
        Time until next health check : 10
        HTTP '404' codes        : 30 (94 total)
        HTTP '500' codes        : 0 (0 total)
        HTTP '502' codes        : 0 (15 total)
Maximum forwarding queue size   : 1000
        Connection 000 queue size       : 0
        Connection 001 queue size       : 0
        Connection 002 queue size       : 0

First ten lines of the file display APM stats (see Application performance monitoring page).
Next two lines show number of pinned sessions and total schedulable connections count for current server.
Next line display server's health monitor. Value 0 means that health monitoring is disabled. If health monitoring is enabled for current server (value 1) then next several lines are displayed with more detailed health information:

• HTTP availability of server (1 - available, 0 - not available).
• Time in seconds until next monitoring request will be sent to server.
• Lines for each tracked HTTP status with number of accumulated responses for period, specified in server_failover_http, and also for the entire period, starting from the Tempesta launch (total). For more information refer to health monitor page.

And last lines display maximum size of forwarding queue with current size for each server connection.