Wordpress to jekyll
30 January 2017 · Filed in DevelopmentPublishing a blog with Jekyll
I had an idea. Why not publish the blog using jekyll and host it on AWS S3? Working with Puppet and ruby, I’m already very familiar with gems and getting jekyll working on my windows 10 workstation with RubyMine was relatively easy.
As this source code is on my github repository, I would have gone with the easy Github Pages option. So you will find this site on (https://neilmillard.github.io/) but I wanted to learn something new.
Hosting on S3
S3 has a feature to host static websites direct from S3, serverless you could say. With this in mind it was quite easy to create a new bucket with the name of the domain and turn on this option.
Next I uploaded the _site directory to the bucket after compiling the site with jekyll build
Migrating the blog
I’m not the first person to take this route and as such there are blog migration tools available to ingest my wordpress blog and create a bunch of posts from it. This worked quite well, but I did have to tweak some of the html for paragraph tags.
Automating the build
Now I would not be an awesome DevOps dude if I didn’t want to automate the build and upload to S3. AWS have been adding services throughout last year and one of these tools is AWS CodeBuild.
Error: invalid byte sequence in US-ASCII
The building of the site was the easy bit, with the exception of adding HTML Proofer, which didn’t want to run due to a UTF-8 argument with US-ASCII. It appears that the default Ruby Docker container is configured with US-ASCII, which really breaks some stuff. With AWS CodeBuild you specify a build file buildspec.yml with which you can configure certain settings, including environment variables.
environment_variables:
plaintext:
LC_ALL: C.UTF-8
ENV LANG: en_US.UTF-8
ENV LANGUAGE: en_US.UTF-8
With these in place HTML Proofer was happy enough and spotted the <p> tags I mentioned earlier.
AWS CodeBuild says it will upload the compiled artifacts to S3. ‘Great this should be easy’ I thought. One minor snag, it needs a key to upload to. This means either a tarball or a subfolder, neither in the format I needed for the website to be hosted correctly. Again open source community to the rescue, with Stout a tool to deploy static websites to S3.
After a few more minor changes, like choosing a theme and adding font-awesome, I’m happy with how it turned out.
Adding a job to AWS CodePipeline, by pointing the source step at the Github repo (after giving permission) I have a two step Pipeline to fetch code, then build it. with the deploy step already in the build step.
Previous Post: AWS announce London region Next Post: The Importance of testing backups Tags: building · jekyll