Justin Duke

Deploying Hugo to S3 with s3cmd

This site (as well as my personal one) is built on Hugo and hosted on S3. I love this combination for a number of reasons:

  1. It’s incredibly fast to build. (This site, as I’m writing it, builds in 26 ms. Milliseconds. That’s insane.)
  2. It’s incredibly cheap. My S3 hosting costs are less than $5 a month.
  3. Deployment is easy. Each post is a flat .md file, and it takes two CLI commands to build and deploy the site.

Let’s talk about that third one a bit, because S3 gets an unfair amount of flak for being a little rough around the edges.

As far as I’m concerned, the S3 tool to end all S3 tools is s3cmd – it’s free, it’s been around forever, and it’s incredibly powerful.

Once you download it and set it up, the command to sync a folder (in this case the public/ folder generated by Hugo once you build the site) with an S3 bucket is actually pretty simple:

s3cmd sync ./ s3://villageblacksmith.consulting --acl-public --recursive --skip-existing

There’s a bunch of stuff going on here:

  • We specify the current folder to sync with ./, rather than having it be implicit, to speed up the operation.
  • We specify the destination bucket with s3://villageblacksmith.consulting. (villageblacksmith.consulting is my bucket name, so just replace that with whatever yours is.)
  • We specify --acl-public so that all of the uploaded files are accessible to the public.
  • We specify --recursive so that it syncs everything in that folder.
  • We specify --skip-existing so that we don’t unnecessarily reupload files that haven’t changed (this is one of the coolest parts of the entire process – it means that when I write a new post I don’t have to reupload the entire site, just a couple pages!)

But that’s it. You should be able to take this command and use it for your Hugo deployment as well. Have fun!

Liked this post? You should subscribe to my newsletter and follow me on Twitter.

(I've got an RSS feed, too, if you'd prefer.)