Cryptophobia / go-monitor

go-monitor is a tool that collects metrics on external urls and exposes metrics for Prometheus to track.

go-monitor

go-monitor is a tool that collects metrics on external urls and exposes metrics for Prometheus to track.

go-monitor uses:

• Gorilla web toolkit mux for the routing.
• Prometheus time-series database for collecting metrics/monitoring.
• prometheus/client_golang for registering collections of custom metrics for the Prometheus metrics collector.
• tcnksm/go-httpstat for precise timing of the go client http request.
• go-http-metrics for collecting metrics about API routes.
• docker-compose for building, developing, and testing locally.

Configuration:

go-monitor needs a config.json file with this minimal structure:

{
  "URLs": [
    {
      "name": "VMware",
      "url": "https://www.vmware.com"
    }
  ]
}

Defaults for CheckTimer options (in seconds) are:

"CheckTimer": {
  "interval": 30,
  "timeout": 5
}

You can also specify the config file path with the -config flag. For example:

go run main.go -config a/custom/path/my_config.json

Kubernetes pod command:

command: go run main.go -config /go/src/app/config.json

Metrics exported:

Name:      "sample_external_url_response_ms",
Subsystem: "go_monitor",
Help:      "Response time in ms for the external url",
Buckets:   []float64{50, 75, 100, 125, 150, 175, 200, 225, 250, 275, 300, 325, 350, 375, 400, 500, 1000, 2000}
Type:      Histogram

Name:      "sample_external_url_up",
Subsystem: "go_monitor",
Help:      "Monitor URL: 0 if the url is down, 1 if the url is up",
Type:      Gauge

Run go-monitor locally:

make start TAG=latest ENV=dev

Builds the go-monitor and then runs it in the background - http://localhost:5000/metrics is the only route exposed for prometheus. Root http://localhost:5000/ is also exposed with a status page and version information.

To access Prometheus UI for metrics and queries: http://localhost:9090/

Testing go-monitor locally:

go test ./monitor/tests/...

Tearing down go-monitor and prometheus:

make stop TAG=latest ENV=dev

Images of Sample Metrics Prometheus UI Queries:

Deploying to Kubernetes Cluster:

Check out the sample Kubernetes kubectl deployment manifests in ./k8s folder.

Original Problem:

Overview

Create a solution (in either Golang or Python) designed to run on a Kubernetes Cluster to monitor internet urls and provide Prometheus metrics, once completed please upload your solution to your github.com account.

Requirements

A service written in python or golang that queries 2 urls (https://httpstat.us/503 & https://httpstat.us/200)
The service will check the external urls (https://httpstat.us/503 & https://httpstat.us/200 ) are up (based on http status code 200) and response time in milliseconds
The service will run a simple http service that produces  metrics (on /metrics) and output a prometheus format when hitting the service /metrics url

    Expected response format:

        sample_external_url_up{url="https://httpstat.us/503 "}  = 0
        sample_external_url_response_ms{url="https://httpstat.us/503 "}  = [value]
        sample_external_url_up{url="https://httpstat.us/200 "}  = 1
        sample_external_url_response_ms{url="https://httpstat.us/200 "}  = [value]

Looking for:

    Code in python or golang
    Dockerfile to build image
    Kubernetes Deployment Specification to deploy Image to Kubernetes Cluster
    Unit Tests
    Good readme providing instructions for use
    Points awarded for screen shots where the metrics are ingested by Prometheus Server

Artifacts should be uploaded to github.com