Hosting static HTML pages generated by Jekyll or any other static site generator with pretty permalinks (the one’s without .html) on to S3 would yield into a 404. The reason is that S3 is an object store hence it doesn’t really look up for a .html version of a page’s permalink.

The solution is copying the file to S3 without the .html extension and explicitly setting the content type of the file, for example:

1

aws s3 cp index.html s3://bucket/index --content-type 'text/html'

By this S3 would send the file contents along with the content-type header set to text/html. To scale, I use the below script which automates this:

1
2
3
4
5
6

# In your public or _site directory
aws s3 cp ./ s3://bucket/ --recursive --exclude "*.html"

for file in $(find . -name '*.html' | sed 's|^\./||'); do
    aws s3 cp ${file%} s3://bucket/${file%.*} --content-type 'text/html'
done

Here’s what it does:

1. Push all the files to S3 excluding all .html files
2. Iterate all the .html files and push them to S3 without the extension and with content-type