dogstatsd

dogstatsd

Name

dogstatsd - publish coredns metrics to dogstatsd agents

Description

CoreDNS has a builtin mechanism for exposing metrics to Prometheus over a HTTP endpoint, but infrastructures that use a metric collection system based on dogstatsd have to run a bridge to poll CoreDNS's metrics endpoint and push them to a dogstatsd agent.

The dogstatsd plugin removes the need for a such bridge by allowing CoreDNS to directly push its metrics to a dogstatsd agent over UDP.

Syntax

dogstatsd [ADDR:PORT]

• ADDR Address at which a dogstatsd agent is available. It may be prefixed with udp://, udp4://, udp6://, or unixgram:// to indicate the protocal to use to push metrics to a dogstatsd agent. If unxigram:// is specified the address must be a path to a unix domain socket on the file system.
• PORT Port number at which the dogstatsd agent is accepting metrics. The port must not be set when the unixgram:// protocol is used to push metrics to a dogstatsd agent.

If you want more control:

dogstatsd [ADDR:PORT] {
    buffer SIZE
    flush INTERVAL
    go
    process
}

• buffer configures the size of the client buffer used to push metrics to a dogstatsd agent. This must not exceed the size of the receive buffer used by the agent. The minimum size is 512 B, the maximum is 64 KB.
• flush configures the time interval between flushes of metrics to a dogstatsd agent. The minimum interval is 1 second, there is not maximum.
• go enables reporting of go metrics to the dogstatsd agent.
• process enables reporting of process metrics to the dogstatsd agent.

Examples

Enable the dogstatsd plugin with a client buffer size of 8 KB, and flushing metrics every 10 seconds.

. {
    dogstatsd localhost:8125 {
        buffer 8192
        flush 10s
    }
}

plugins.cfg

This plugin is intended to appear right after the prometheus plugin declaration.