As I move forward with my Beego project, I have reached a point where I need to authenticate my users. In the flow that I’m looking for, the client (web app, mobile app, etc…), will communicate directly with the Auth provider (Google, Facebook, etc…) and get a JWT. The only thing the server needs to do is validate the JWT. If the validation succeeds, it means the user is logged in.

If you are not familiar with JWT, you can read my previous article that explains how JWT works.

Authentication flow

Since the authentication with Google is going to happen entirely on the client, the server logic becomes a lot simpler. For my application, all endpoints will require the user to be logged in, so I will create a middleware to verify this. The middleware will expect a valid JWT in the Authorization header. If this requirement is not met, the server will return a 401.

Before we start creating models, we need to make sure our app can connect to our database. At the time of this writing Beego supports MySQL, PostgreSQL and Sqlite3.

We can have the ORM connect to our database from our main.go file:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
package main

import (
    "os"
    "strings"

    _ "app/routers"

    "github.com/astaxie/beego"
    "github.com/astaxie/beego/orm"
    _ "github.com/go-sql-driver/mysql"
)

func init() {
    orm.RegisterDriver("mysql", orm.DRMySQL)
    parts := []string{os.Getenv("MYSQL_USER"), ":", os.Getenv("MYSQL_PASSWORD"),
            "@tcp(", os.Getenv("MYSQL_HOST"), ":3306)/", os.Getenv("MYSQL_DATABASE")}
    orm.RegisterDataBase("default", "mysql", strings.Join(parts, ""))
}

func main() {
    if beego.BConfig.RunMode == "dev" {
        beego.BConfig.WebConfig.DirectoryIndex = true
        beego.BConfig.WebConfig.StaticDir["/swagger"] = "swagger"
    }
    beego.Run()
}

A few weeks ago I wrote an article showing how to start a simple Beego server. Today I’m going to go a little further and explain how to start building your database using migrations. For my examples I will use a MySQL server with an empty database (no tables) named beego.

It is possible to just log in to a database server and type the SQL commands necessary to create or modify tables when we need to, but using migrations helps keep track of the changes made to the database over time. This provides something similar to version control at the database level.

If we need to create a new table, we would start by creating a new migration file using the bee tool:

1
bee generate migration create_user_table

I tried GoLang for the first time a little less than two years ago and I didn’t get a very good first impression of it. One of the things that I disliked the most is that back then the community seemed to be overly focused in performance even when that meant creating unreadable code.

One of the things that bothered me the most was that the community was against creating web frameworks with opinions. Because of this lack of standardization, I had to relearn how things are done every time I looked at a new web project written in Go.

Today I decided to look again into the state of web frameworks and there seems to be a few promising options out there. I saw a good amount of people recommending Beego, so I decided to give it a try.

It has been more than a year since I wrote an article about dependency management with Glide. It seems like things have changed a little since then. An official package manager has been started by the community, which hopefully will make things easier for developers.

Install

For go applications more than with any other language (because of the necessity of GOPATH), I highly recommend using docker. This is a minimal Dockerfile I’m using that includes both Go and Dep:

In a previous post I explored using Bitbucket Pipelines to generate and publish Docker images for my projects. I was worried I would reach the 50 minutes limit pretty quickly, but even before I reached it I ran into other issues that made me look for other options. Namely, you can’t use docker-compose or the docker run command in Pipelines.

While looking for other options I found GitLab. I had heard a lot of things about GitLab in the past, especially about their very advanced deployment pipelines capability. One thing that I didn’t know, which makes me very happy, is that they have a free tier that includes 2,000 minutes of CI per month (Around 1 hour per day). This number should be more than enough for my personal project needs.

I have a few projects that I host on Bitbucket (Mostly because I can have private repos for free). As I was working on some of these projects last week, I realized that there are a lot of manual steps I have to execute in order to verify that my project is in good health and to publish it or deploy it.

Today I’m going to explore using Bitbucket’s Pipelines to generate a Docker image out of one of my projects and publish it to Canister.

I’m trying to get a little Rails application ready for production, and just for fun I’m trying to make the image a little slimmer than it currently is (860 MB).

There is a official docker image of ruby with alpine (ruby:alpine), but because of the libraries that rails uses (with native bindings), it is a little more challenging than just referencing that image on the top of the Dockerfile.

Solving the issues

I added FROM ruby:2.4.1-alpine at the top of my Dockerfile and tried to create the image. The first problem I faced was with myql2. For the mysql2 gem to work on Alpine it is necessary to have a compiler (build-base) and MySQL development libraries (mariadb-dev). I added this to my Dockerfile:

I was doing a little studying on algorithms when I stumbled into what looked to me as a pretty simple question: “Find the first occurrence of a string inside another string”. This can be simply achieved with two nested loops and a worst case scenario performance of O(nm) or O(n^2) if m’s size is relative to n.

Here is an implementation that uses two loops:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
function findString(s1, s2) {
  // We only need to loop until s2 doesn't fit anymore
  var loopLimit = s1.length - s2.length;
  for (var i = 0; i <= loopLimit; i++) {
    for (var j = 0; j < s2.length; j++) {
      // As soon as we find a mismatch we abort the loop
      if (s1[i + j] !== s2[j]) {
        break;
      }
    }

    // If j reached the size of s2, it means all letters
    // in s2 matched. We have succeeded!
    if (j === s2.length) {
      return true;
    }
  }

  return false;
}

findString('aaaaaaaaaa', 'aaaab'); // false
findString('aaaaaaaaab', 'aaaab'); // true
findString('aaaaaaaaab', 'ab'); // true

I’m building a very simple app using Rails. While looking for guidance for preparing it for production, I found a lot of articles suggesting to put Nginx in front of it. After talking to some people they explained some reasons why this is suggested:

• Ngingx can serve static assets – This appears to be the greatest and clearer advantage. You can configure Ngingx to directly serve static assets without having to hit Rails at all. This is very good because every request that comes to Rails will block all other request because Ruby is single threaded
• Nginx can do caching for you – Nginx can cache some of the static assets, which would give them a performance boost
• Nginx is multithreaded – Nginx can serve multiple static assets at the same time Rails is serving requests

These are definitely advantages (specially the first one), but having Nginx in front of my server also adds complexity to my deployment. To figure out if the added complexity worth it, I decided to run some load tests. Here I will explain how I did it and what were the results.