s3deploy

A simple tool to deploy static websites to Amazon S3 and CloudFront with Gzip and custom headers support (e.g. "Cache-Control"). It uses ETag hashes to check if a file has changed, which makes it optimal in combination with static site generators like Hugo.

• Install
• Configuration
  • Flags
  • Routes
• Global AWS Configuration
• Example IAM Policy
• CloudFront CDN Cache Invalidation
  • Example IAM Policy With CloudFront Config
• Background Information
• Alternatives
• Stargazers over time

Install

Pre-built binaries can be found here.

s3deploy is a Go application, so you can also install the latest version with:

 go install github.com/bep/s3deploy/v2@latest

To install on MacOS using Homebrew:

brew install bep/tap/s3deploy

Note The brew tap above currently stops at v2.8.1; see this issue for more info.

Note that s3deploy is a perfect tool to use with a continuous integration tool such as CircleCI. See this for a tutorial that uses s3deploy with CircleCI.

Configuration

Flags

The list of flags from running s3deploy -h:

-V	print version and exit
-acl string
    provide an ACL for uploaded objects. to make objects public, set to 'public-read'. all possible values are listed here: https://docs.aws.amazon.com/AmazonS3/latest/userguide/acl-overview.html#canned-acl (default "private")
-bucket string
    destination bucket name on AWS
-config string
    optional config file (default ".s3deploy.yml")
-distribution-id value
    optional CDN distribution ID for cache invalidation, repeat flag for multiple distributions
-endpoint-url string
    optional endpoint URL
-force
    upload even if the etags match
-h	help
-ignore value
    regexp pattern for ignoring files, repeat flag for multiple patterns,
-key string
    access key ID for AWS
-max-delete int
    maximum number of files to delete per deploy (default 256)
-path string
    optional bucket sub path
-public-access
    DEPRECATED: please set -acl='public-read'
-quiet
    enable silent mode
-region string
    name of AWS region
-secret string
    secret access key for AWS
-skip-local-dirs value
    regexp pattern of files of directories to ignore when walking the local directory, repeat flag for multiple patterns, default "^\\/?(?:\\w+\\/)*(\\.\\w+)"
-skip-local-files value
    regexp pattern of files to ignore when walking the local directory, repeat flag for multiple patterns, default "^(.*/)?/?.DS_Store$"
-source string
    path of files to upload (default ".")
-strip-index-html
    strip index.html from all directories expect for the root entry
-try
    trial run, no remote updates
-v	enable verbose logging
-workers int
    number of workers to upload files (default -1)

The flags can be set in one of (in priority order):

1. As a flag, e.g. s3deploy -path public/
2. As an OS environment variable prefixed with S3DEPLOY_, e.g. S3DEPLOY_PATH="public/".
3. As a key/value in .s3deploy.yml, e.g. path: "public/"
4. For key and secret resolution, the OS environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY (and AWS_SESSION_TOKEN) will also be checked. This way you don't need to do any special to make it work with AWS Vault and similar tools.

Environment variable expressions in .s3deploy.yml on the form ${VAR} will be expanded before it's parsed:

path: "${MYVARS_PATH}"
max-delete: "${MYVARS_MAX_DELETE@U}"

Note the special @U (Unquoute) syntax for the int field.

Skip local files and directories

The options -skip-local-dirs and -skip-local-files will match against a relative path from the source directory with Unix-style path separators. The source directory is represented by ., the rest starts with a /.

Strip index.html

The option -strip-index-html strips index.html from all directories expect for the root entry. This matches the option with (almost) same name in hugo deploy. This simplifies the cloud configuration needed for some use cases, such as CloudFront distributions with S3 bucket origins. See this PR for more information.

Routes

The .s3deploy.yml configuration file can also contain one or more routes. A route matches files given a regexp. Each route can apply:

header : Header values, the most notable is probably Cache-Control. Note that the list of system-defined metadata that S3 currently supports and returns as HTTP headers when hosting a static site is very short. If you have more advanced requirements (e.g. security headers), see this comment.

gzip : Set to true to gzip the content when stored in S3. This will also set the correct Content-Encoding when fetching the object from S3.

Example:

routes:
    - route: "^.+\\.(js|css|svg|ttf)$"
      #  cache static assets for 1 year.
      headers:
         Cache-Control: "max-age=31536000, no-transform, public"
      gzip: true
    - route: "^.+\\.(png|jpg)$"
      headers:
         Cache-Control: "max-age=31536000, no-transform, public"
      gzip: false
    - route: "^.+\\.(html|xml|json)$"
      gzip: true

Global AWS Configuration

See https://docs.aws.amazon.com/sdk-for-go/api/aws/session/#hdr-Sessions_from_Shared_Config

The AWS SDK will fall back to credentials from ~/.aws/credentials.

If you set the AWS_SDK_LOAD_CONFIG environment variable, it will also load shared config from ~/.aws/config where you can set the global region to use if not provided etc.

Example IAM Policy

{
   "Version": "2012-10-17",
   "Statement":[
      {
         "Effect":"Allow",
         "Action":[
            "s3:ListBucket",
            "s3:GetBucketLocation"
         ],
         "Resource":"arn:aws:s3:::<bucketname>"
      },
      {
         "Effect":"Allow",
         "Action":[
            "s3:PutObject",
            "s3:PutObjectAcl",
            "s3:DeleteObject"
         ],
         "Resource":"arn:aws:s3:::<bucketname>/*"
      }
   ]
}

Replace with your own.

CloudFront CDN Cache Invalidation

If you have configured CloudFront CDN in front of your S3 bucket, you can supply the distribution-id as a flag. This will make sure to invalidate the cache for the updated files after the deployment to S3. Note that the AWS user must have the needed access rights.

Note that CloudFront allows 1,000 paths per month at no charge, so S3deploy tries to be smart about the invalidation strategy; we try to reduce the number of paths to 8. If that isn't possible, we will fall back to a full invalidation, e.g. "/*".

Example IAM Policy With CloudFront Config

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "s3:ListBucket",
                "s3:GetBucketLocation"
            ],
            "Resource": "arn:aws:s3:::<bucketname>"
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:PutObject",
                "s3:DeleteObject",
                "s3:PutObjectAcl"
            ],
            "Resource": "arn:aws:s3:::<bucketname>/*"
        },
        {
            "Effect": "Allow",
            "Action": [
                "cloudfront:GetDistribution",
                "cloudfront:CreateInvalidation"
            ],
            "Resource": "*"
        }
    ]
}

Background Information

If you're looking at s3deploy then you've probably already seen the aws s3 sync command - this command has a sync-strategy that is not optimised for static sites, it compares the timestamp and size of your files to decide whether to upload the file.

Because static-site generators can recreate every file (even if identical) the timestamp is updated and thus aws s3 sync will needlessly upload every single file. s3deploy on the other hand checks the etag hash to check for actual changes, and uses that instead.

Alternatives

• go3up by Alexandru Ungur
• s3up by Nathan Youngman (the starting-point of this project)

Stargazers over time