Statsd Module

Eloy Coto Pereiro

   <eloy.coto@gmail.com>

Edited by

Eloy Coto Pereiro

   <eloy.coto@gmail.com>

   Copyright © 2014 Eloy Coto
     __________________________________________________________________

   Table of Contents

   1. Admin Guide

        1. Overview
        2. Parameters

              2.1. ip(string)
              2.2. port(string)

        3. Functions

              3.1. statsd_set(key, value[, labels])
              3.2. statsd_gauge(key, value[, labels])
              3.3. statsd_histogram(key, value[, labels])
              3.4. statsd_start(key)
              3.5. statsd_stop(key[, labels])
              3.6. statsd_incr(key[, labels])
              3.7. statsd_decr(key[, labels])

        4. KEMI bindings

   List of Examples

   1.1. Set ip parameter
   1.2. Set port parameter
   1.3. statsd_set usage
   1.4. statsd_set usage with labels
   1.5. statsd_gauge usage
   1.6. statsd_gauge usage with labels
   1.7. statsd_histogram usage
   1.8. statsd_histogram usage with labels
   1.9. statsd_start usage
   1.10. statsd_stop usage
   1.11. statsd_stop usage with labels
   1.12. statsd_incr usage
   1.13. statsd_incr usage with labels
   1.14. statsd_decr usage
   1.15. statsd_decr usage with labels

Chapter 1. Admin Guide

   Table of Contents

   1. Overview
   2. Parameters

        2.1. ip(string)
        2.2. port(string)

   3. Functions

        3.1. statsd_set(key, value[, labels])
        3.2. statsd_gauge(key, value[, labels])
        3.3. statsd_histogram(key, value[, labels])
        3.4. statsd_start(key)
        3.5. statsd_stop(key[, labels])
        3.6. statsd_incr(key[, labels])
        3.7. statsd_decr(key[, labels])

   4. KEMI bindings

1. Overview

   The module implements the statsd text protocol, allowing Kamailio to
   emit counters, gauges, histograms and timers to any server that
   understands statsd semantics. Typical destinations include the
   reference Etsy/Graphite daemon, InfluxDB via Telegraf listeners, or any
   other statsd-compatible collector.

   The implementation has no special external dependencies; it opens a UDP
   socket to the configured host and sends the raw statsd datagrams, so
   any standards-compliant server can receive them.

2. Parameters

   2.1. ip(string)
   2.2. port(string)

2.1. ip(string)

   Statsd server IP address.

   Defaults to 127.0.0.1

   Example 1.1. Set ip parameter
...
modparam("statsd", "ip", "127.0.0.1")
...

2.2. port(string)

   Statsd server port number

   Defaults to 8125

   Example 1.2. Set port parameter
...
modparam("statsd", "port", "8125")
...

3. Functions

   3.1. statsd_set(key, value[, labels])
   3.2. statsd_gauge(key, value[, labels])
   3.3. statsd_histogram(key, value[, labels])
   3.4. statsd_start(key)
   3.5. statsd_stop(key[, labels])
   3.6. statsd_incr(key[, labels])
   3.7. statsd_decr(key[, labels])

3.1.  statsd_set(key, value[, labels])

   Sets count the number of unique values passed to a key.

   If this method is called multiple times with the same userid in the
   same sample period, that userid will only be counted once.

   Optional labels supply a comma-separated list of key:value tags; omit
   the argument to emit the metric untagged.

   This function can be used in ALL ROUTES.

   Example 1.3. statsd_set usage
...
failure_route[tryagain] {
...
    statsd_set("customerFailure", 1);
...
}
...

   Example 1.4. statsd_set usage with labels
...
statsd_set("active_users", $var(count), "cluster:a,region:$var(region)");
...

3.2.  statsd_gauge(key, value[, labels])

   Gauges are a constant data type. They are not subject to averaging and
   they don't change unless you change them. That is, once you set a gauge
   value, it will be a flat line on the graph until you change it again.

   Gauges are useful for things that are already averaged, or don't need
   to reset periodically

   This function can be used in ALL ROUTES.

   The statsd server collects gauges under the stats.gauges prefix.
   Optional labels supply a comma-separated list of key:value tags; omit
   the argument to emit the metric untagged.

   Example 1.5. statsd_gauge usage
...
route [gauge_method]{
    statsd_gauge("method"+$rm, "+1");
    statsd_gauge("customer_credit"+$var(customer),"$var(customer_credit)");
}
...

   Example 1.6. statsd_gauge usage with labels
...
statsd_gauge("method"+$rm, "+1", "node:$si");
...

3.3.  statsd_histogram(key, value[, labels])

   The histograms are a measure of time, but they are calculated at the
   server side. As the data exported by the client is the same, this is
   just an alias for the Timer type.

   This function can be used in ALL ROUTES.

   The statsd server collects histograms under the stats.histograms
   prefix. Optional labels supply a comma-separated list of key:value
   tags; omit the argument to emit the metric untagged.

   Example 1.7. statsd_histogram usage
...
    statsd_histogram("latency", 1000);
...

   Example 1.8. statsd_histogram usage with labels
...
statsd_histogram("latency", 1000, "method:"+$rm+",peer:"+$si);
...

3.4.  statsd_start(key)

   statsd_start set an avp with the key name, and when statsd_stop(key) is
   used, the module will send statsd the difference in milliseconds. This
   is useful to know the time of a SQL query, or how much time your
   replies take.

   This function can be used in all routes.

   The statsd server collects all timers under the stats.timers prefix and
   will calculate the lower bound, mean, 90th percentile, upper bound, and
   count of each timer for each period (by the time it can be seen in
   graphite, that's usually per minute).

   Example 1.9. statsd_start usage
...
statsd_start("long_mysql_query");
sql_query("ca", "select sleep(0.2)", "ra");
statsd_stop("long_mysql_query");
...

3.5.  statsd_stop(key[, labels])

   statsd_stop(key) get the avp string with the key and calculate the
   difference from the start time. When finished the milliseconds used
   will be sent to statsd.

   Optional labels supply a comma-separated list of key:value tags; omit
   the argument to emit the metric untagged.

   This function can be used in all routes.

   Example 1.10. statsd_stop usage
...
statsd_start("long_mysql_query");
sql_query("ca", "select sleep(0.2)", "ra");
statsd_stop("long_mysql_query");
...

   Example 1.11. statsd_stop usage with labels
...
statsd_start("long_mysql_query");
sql_query("ca", "select sleep(0.2)", "ra");
statsd_stop("long_mysql_query", "db:replica,shard:$var(shard)");
...

3.6.  statsd_incr(key[, labels])

   Increment a statsd counter

   This function can be used in all routes.

   Optional labels supply a comma-separated list of key:value tags; omit
   the argument to emit the metric untagged.

   Example 1.12. statsd_incr usage
...
if(geoip_match("$si", "src")){
    statsd_incr("country."+$(gip(src=>cc)));
}
...

   Example 1.13. statsd_incr usage with labels
...
if(geoip_match("$si", "src")){
        statsd_incr("country."+$(gip(src=>cc)), "asn:"+$(gip(src=>asn)));
}
...

3.7.  statsd_decr(key[, labels])

   Decrement a counter

   This function can be used in all routes.

   Example 1.14. statsd_decr usage
...
if (t_check_status("408")) {
    statsd_decr("kamailio.successfulCalls");
    statsd_incr("kamailio.reply.timeout");
}
...

   Example 1.15. statsd_decr usage with labels
...
if (t_check_status("408")) {
        statsd_decr("kamailio.successfulCalls", "response:timeout");
        statsd_incr("kamailio.reply.timeout", "response:timeout");
}
...

4. KEMI bindings

   All of the functionality above is also exported to Kamailio's scripting
   interfaces (Lua, Python, JS, etc.) via the statsd KEMI module. For
   every function that accepts labels there is a companion name suffixed
   with _labels. For example, statsd_gauge and statsd_labels_gauge,
   statsd_histogram and statsd_labels_histogram, statsd_set and
   statsd_labels_set, statsd_incr and statsd_labels_incr, and so on.
   Timers are exposed as statsd_start, statsd_stop and statsd_labels_stop.
