#prometheus-metrics #connector #format #http #retrieve #internals #sōzu

app sozu-prometheus-connector

This application retrieve internals metrics of Sōzu and format them into prometheus

4 releases ()

1.1.0-rc.0 Oct 21, 2024
0.2.3 Oct 21, 2024
0.2.2-rc.1 Jul 25, 2024
0.2.2-rc.0 Jul 19, 2024

#352 in Network programming

Download history 138/week @ 2024-07-23 16/week @ 2024-07-30 18/week @ 2024-09-10 8/week @ 2024-09-17 16/week @ 2024-09-24 32/week @ 2024-10-01 163/week @ 2024-10-15 117/week @ 2024-10-22 2/week @ 2024-10-29

282 downloads per month

Custom license and LGPL-3.0

42KB
824 lines

Sōzu Prometheus connector

Receives HTTP GET requests on a /metrics route, forwards the request to Sōzu, packages the responses in a Prometheus format and sends them back in a HTTP response.

Configure

In the config.toml of this repository, you have to indicate the absolute path to the configuration file of the Sōzu that runs on the machine.

sozu_configuration_path = "/path/to/sozu/on/the/machine/config.toml"

The Sōzu config file will be parsed to find to unix socket on which to write requests to Sōzu.

You will have to provide a socket address on which the prometheus connector will wait for HTTP requests on the /metrics path.

# address on which to listen. Must be parsable to the SocketAddr type
listening_address = "0.0.0.0:3000"

you can also chose to flatten all metric and to compound them if they have the same name, turning thig:

# TYPE bytes_out counter
bytes_out{worker="0"} 246
bytes_out{cluster_id="MyCluster",backend_id="the-backend-to-my-app"} 250
bytes_out{cluster_id="MyCluster",backend_id="the-backend-to-my-app-2"} 250
bytes_out{worker="1"} 246
bytes_out{cluster_id="MyCluster",backend_id="the-backend-to-my-app"} 250
bytes_out{cluster_id="MyCluster",backend_id="the-backend-to-my-app-2"} 250

into this:

# TYPE bytes_out counter
bytes_out{} 492
bytes_out{cluster_id="MyCluster"} 1000

You can do this with:

aggregate-backend-metrics = true

How to test

  1. Run Sōzu on your machine
  2. Run sozu-prometheus-connector with cargo run -- --config config.toml
  3. In a web browser, query the URL 127.0.0.1:3000

The prometheus-formatted metrics should appear in the browser.

Dependencies

~33–67MB
~1M SLoC