adak

README

Adak

Adak is an e-commerce server developed for educational purposes.

This project has two versions: monolithic and microservices.

For the microservices version please change to the microservices branch.

Features

• Cookie-based sessions encrypted using ChaCha20-Poly1305
• Basic authentication and OAuth2 (Google)
• Password encryption using bcrypt
• Email and admins verification
• OpenAPI Specification 3.0.0 with Swagger
• Stripe integration for payments
• Pagination, caching, rate limiting, GZIP responses compression, input sanitization and validation, context cancelling

Installation

Docker

Optional: mount your configuration files and certificates in the docker-compose.yml file, or remove them to use the default configuration.

docker-compose up

Manually

Requires postgres, memcached, redis, Go and Git to be installed

Clone the repository:

git clone https://www.github.com/GGP1/adak.git

Configuration: set an environment variable called ADAK_CONFIG pointing to your configuration file.

Run the server:

go run cmd/main.go
# or
make run

Documentation

API

The API's documentation can be found on SwaggerHub.

Database migrations

The golang-migrate/migrate CLI is required to execute the commands.

Up:

migrate -path pkg/postgres/migrations -database <postgres_url> up

Down:

migrate -path pkg/postgres/migrations -database <postgres_url> down

Pagination

Adak accepts bot limit and cursor url parameters for paginating results.

The next_cursor field is always included in the response if there are more values to be fetched. The next cursor is builded up from the last record's created_at timestamp and uuid, they are formatted and concatenated, the resulting string is encoded with base64 and sent to the client.

Example response:

{
"next_cursor": "MjAyMS0wNy0wOVQyMzo0ODozOC42NDg0MDZaLHhLM0lBY01XenFTdmY5aEdyUVpsOENOc1haeUV2NlBJ",
    "users": [
        {
            "id": "xK3IAcMWzqSvf9hGrQZl8CNsXZyEv6PI",
            "created_at": "2021-07-09T23:48:38.648406Z",
        }
    ]
}

Sample request:

localhost:4000/users?limit=5&cursor=MjAyMS0wNy0wOVQyMzo0ODozOC42NDg0MDZaLHhLM0lBY01XenFTdmY5aEdyUVpsOENOc1haeUV2NlBJ

Amounts

Amounts are represented by 64-bit integers to be provided in a currency's smallest unit (100 = 1 USD).

In the case of weights, 1000 = 1kg.

Monitoring

Adak collects information using prometheus and runs a grafana container for visualizing it.

The data is gathered from multiple sources:

• Server: requests and database hits
• Node: hardware and OS metrics provided by node exporter

More data can be extracted from other services like Postgres, Redis and Memcached, although they are not implemented, it would imply configuration changes only.

Visualizing metrics

Raw

The server's raw statistics are shown in the "/metrics" path (it also contains Go metrics but not information from external sources).

Prometheus

Prometheus allows not only querying the data but also displaying it in graphs, this can be done at http://localhost:9090 when running with docker compose.

Grafana

To display prometheus' metrics with grafana follow these steps:

1. Go to http://localhost:3000 and log in as "admin", the password is "admin".
2. Add prometheus as a data source. Set HTTP/Access to Browser and the url to http://localhost:9090, then click Save & Test.
3. On the top of the page and next to the Settings tab there other called Dashboards, click it and import Prometheus 2.0 Stats.
4. Finally, go to the dashboards home and select the dashboard imported above.

You can create your own or use official/community built dashboards.

Load testing

Making 10000 requests with 100 concurrent workers to the /users endpoint, each response containing 20 non-cached user objects (2.94KB per response) resulted in:

$ hey -n 10000 -c 100 -m GET http://:4000/users

Summary:
  Total:        11.4822 secs
  Slowest:      0.4250 secs
  Fastest:      0.0021 secs
  Average:      0.1110 secs
  Requests/sec: 870.9148

Response time histogram:
  0.002 [1]     |
  0.044 [1708]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.087 [2219]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.129 [2383]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.171 [1834]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.214 [1141]  |■■■■■■■■■■■■■■■■■■■
  0.256 [491]   |■■■■■■■■
  0.298 [169]   |■■■
  0.340 [38]    |■
  0.383 [13]    |
  0.425 [3]     |

Latency distribution:
  10% in 0.0281 secs
  25% in 0.0608 secs
  50% in 0.1049 secs
  75% in 0.1542 secs
  90% in 0.1982 secs
  95% in 0.2284 secs
  99% in 0.2809 secs

Details (average, fastest, slowest):
  DNS+dialup:   0.0000 secs, 0.0021 secs, 0.4250 secs
  DNS-lookup:   0.0000 secs, 0.0000 secs, 0.0000 secs
  req write:    0.0000 secs, 0.0000 secs, 0.0041 secs
  resp wait:    0.1107 secs, 0.0019 secs, 0.4247 secs
  resp read:    0.0002 secs, 0.0000 secs, 0.0064 secs

Status code distribution:
  [200] 10000 responses