Images are a big part of the web, and also the bigest part of the web bloat. Everyday I see sites way too heavy or way too complicated architectures for dealing with images. With @sadok_f we have been working on having a simple server to deal with images. You request an image, give the desired dimensions and compression… and voilà! you get your image back, cached and compressed using the incredible moz-jpeg.
DigitalOcean is not the only option you have, you can do it on Linode, AWS, even on a VPS or a regular server. But for this example, and since I have an account lying around we will use Digital Ocean.
Mise en place, What we’ll need
- A Digital Ocean account (the account itself it’s free)
- flyimage an image conversion/compression microservice project (take a look for now)
- flyimg Dropplet provision script for DigitalOcean.
New DigitalOcean Droplet
Go to our DigitalOcean dashboard and just create a new droplet.
For the dropplet I’m chosing an Ubuntu 16.04 image (but it also worked with a 14.x). Choose the closest region to us or our audience.
Check User data checkbox to have an initial config of users and libraries. We will take advantage of the cloud-config field available, where we can add YAML instructions to provision the server.
#cloud-config users: - name: leopold groups: sudo shell: /bin/bash sudo: ['ALL=(ALL) NOPASSWD:ALL'] package_upgrade: true packages: - git - docker.io runcmd: - wget https://raw.githubusercontent.com/Flyimg/DigitalOcean-provision/master/provision-flyimg-on-ubuntu.sh - chmod 700 provision-flyimg-on-ubuntu.sh && ./provision-flyimg-on-ubuntu.sh
At the time of writing looks like the above (but check the repo for the latest version).
Also add your SSH key in the fields so you can ssh later to the server. I highly reccomend this because the provision script will close access to root and password login, only allowing access to the user “leopold” with your SSH key. If you don’t have an SSH key this still works, but you can’t finetune your server later on.
Finally just put a name to our server (in my case
flyimg) and start the droplet.
Right after the image is created and the server started you can still
ssh to our machine with root. Here you can take a peek at what is going on with the provisioning of the server.
ssh email@example.com # $ logged in the server tail -f /var/log/cloud-init-output.log
Now you can see live what the robots are doing to setup your server. In the end you should see something like this:
Optionally on a separate console window or session you can check that login in works with user leopold.
After that you can start using the server with the IP of the server, somehting like:
http://188.8.131.52. But it would be prettier to have a domain to point it to.
Use a domain instead of an IP
This is simple, you just need to tell you registrar (or DNS server provider) to create a DNS record type A pointing to the IP of your droplet, it should be available in a few minutes… or a few hours depending on the ISP, it requires propagation.
In this case we have the subdomain img.medula.cl pointed to the server form the (mediatemple) control panel.
In dreamhost looks like the following:
Now you can try accessing the domain directly and you get the “Hello from Docker!” message.
Changing the defaults
By default we can fetch images from any URL, resize, compress and serve them right back. That is easy to abuse. We will limit the domains from which the app can request images. Also I want to have a bunch of source images in this same domain.
To restrict the domain we need to edit
config/parameters.yml file and change. For that you need to access your dropplet through SSH, and here is why you needed to add an SSH key at the creation of your dropplet.
The inside the server edit the
flyimg/config/parameters.yml file (vim, nano, pico, whatever you want).
leopold@flyimg:~$ vim flyimg/config/parameters.yml
and add the domain or domains you want
whitelist_domains: - medula.cl - m0.cl
There’s no need to restart the server, we can just test by trying to fetch an image from flick, like this one: https://c2.staticflickr.com/1/362/19293294393_cbde92e205_k.jpg . If we pass it to the server http://img.medula.cl/upload/w_201,h_201,q_90/https://c2.staticflickr.com/1/362/19293294393_cbde92e205_k.jpg it returns a 403.
But if we use this image: http://m0.cl/t/resize-test_760.jpg and pass it to the server: http://img.medula.cl/upload/w_502,h_501,c_1/http://m0.cl/t/resize-test_760.jpg we get a 200 back with the image.
And that is all you ned to have your own little server resizing, compressing and caching your images.
Thaking it further
The next natural step is to make it serve using
https for that you have at least 2 ways:
- Difficult: Configure let’s encrypt for your Ubuntu following a tutorial from DigitalOcean
- Easy: Use cloudflare to serve your images cached. Although you need to give control of you DNSs to Cloudflare.
Even cooler is to have your transformed images stored in a more permanent medium, like Amazon S3 or DigitalOcean block storage. Flyimg makes it simple because it uses flysystem, A virtual file system that lets you use any storage and handle the files in the same way, like an ORM for file storage.
My next step is to have some sort simple system to organize my images, like a file system, but for images and the web only. I upload there, and use flyimg to do all the transformations and caching.