echod

Echod is an HTTP server which echoes whatever JSON a client sent through the API appending "echoed": true. This is a demo project to demonstrate an API built in Go with logs, metrics, unit and integration tests. I decided to also include Envoy as an HTTPS proxy (for demonstration purposes) and Prometheus on the docker-compose.yaml. If you want to run just the service with docker, you can do so with make dockerrun.

Deploying something useless into production, as soon as you can, is the right way to start a new project. It pulls unknown risk forward, opens up parallel streams of work, and establishes good habits. https://blog.thepete.net/blog/2019/10/04/hello-production/

The API consists on the following endpoints:

• POST /api/echo (requires Authentication)
• PUT /api/echo (requires Authentication)
• GET /health
• GET /metrics

Running

Pre-requisites:

• docker
• docker-compose

make run
curl -k https://localhost:10000/health

Note: Always use curl -k or curl --insecure to bypass the SSL validation since it is a self-signed certificate installed on Envoy. For production we would use something like Let's Encrypt and certbot to avoid this issue.

Running the tests (Go is required):

make test

Running the lints (golangci-lint is required):

make lint

Authentication

To use the /api/echo endpoint you must be authenticated using basic auth with one of the following users and password respectively:

user	password
remarkablebob	2rFtex
dextroussheep	vSKjGK
luckyslug	VRsBMW
pettyrabbit	2BpgwH

Example:

curl -k -u remarkablebob:2rFtex -X POST https://localhost:10000/api/echo --data '{"foo": "bar"}'

{"echoed":true,"foo":"bar"}

Documentation

Pre-requisites:

• go
• godoc (you can install by running go get golang.org/x/tools/cmd/godoc)

make doc

Benchmark

Simple performance benchmark running both the server and wrk locally:

$ wrk -t4 -c400 -d30s --script config/wrk/echo.lua http://localhost:3000/api/echo

Running 30s test @ http://localhost:3000/api/echo
  4 threads and 400 connections
  Thread Stats   Avg      Stdev     Max   +/- Stdev
    Latency     3.47ms    1.66ms  53.39ms   84.25%
    Req/Sec    18.04k     1.89k   24.36k    68.92%
  2153668 requests in 30.01s, 384.08MB read
  Socket errors: connect 153, read 19, write 21, timeout 0
Requests/sec:  71764.55
Transfer/sec:     12.80MB

Monitoring

Echod exposes two metrics on Prometheus format:

• http_requests_total: Measures the total http requests received labeled by status code, method and path (Counter).
• http_request_duration_milliseconds: Measures the duration of each http request (Histogram).

Alerts

For more information, check alerts.yaml. You can run make run and access Prometheus on http://localhost:9000/.

Envoy

• InstanceDown: triggers when Envoy is down for more than 1 minute.

Echod

• InstanceDown: triggers when the service is down for more than 5 minutes.
• HighRequestLatency: triggers when the average requests duration is longer than 500ms.
• HttpLatency99Percentile: triggers when 1% of the slowest requests are longer than 1s.
• HttpHigh5xxErrorRate: triggers when 5% of the requests are HTTP 5xx responses.