Grails cdn-asset-pipeline Plugin for Improving Web Site Performance
We’re always looking to get our software to run faster. For web sites, part of that equation is to move static content to a content delivery network (CDN). A proper CDN is designed to serve static content fast, have multiple geographic domains and long cache times. The load on the application server is lessened, giving more bandwidth for application use rather than static content. This post explores using the Grails cdn-asset plugin to copy and use static resources on the Amazon S3 service. Once configured it makes hosting static content on S3 easy.
Note that the Grails plugin discussed here only supports Amazon S3 at the time of this writing, cdn-asset-pipeline version 0.4.1. If you don’t want to use S3 then this post won’t help you, unless you want to contribute to the plugin and add your chosen provider.
Assets Using the asset-pipeline Plugin
The asset-pipeline plugin manages static resources in your Grails application. It does some very nice things like collecting all similar types into a common file (application.js, application.css, etc.), minification, filenames based on content hash, etc. In development mode these files are broken out to enable reloading without restarting the server. In production the assets are combined and optimized.
The asset-pipeline plugin replaces the capability in the resources plugin. The asset-pipeline plugin supports Grails 2.0 and greater and is the default in Grails 2.4.
What we want to do is move the production resources to a CDN (in this case Amazon S3) and have the application point to the CDN for the static resources. This will improve performance for the client and reduce load on the server. The cdn-asset-pipeline plugin will do that for us.
Configuring Amazon S3
You will need an Amazon account to access the AWS services. If you’re new to AWS you’ll get some free space and bandwidth for the first year. Go to the S3 Console. Create a new bucket for your static resources. It’s best to have a separate bucket for security reasons. If you make a mistake you can remove all the files and upload them again, nothing is lost. If your bucket is comprised, you can replace the static resources.
Configuring the permissions gave me the biggest trouble. You will want a separate user account for uploading to the bucket. Definitely don’t use your Amazon “root” account and don’t share accounts. If you’re going to upload from different systems, for example your development environment and a CI/CD environment, create separate users for those as well. You can then revoke access from a particular system if needed. In the AWS Console, click IAM. If you haven’t used IAM it will step you through a few things to secure your account. After configuring IAM, follow these steps to create new users and give them upload access to your bucket.
- Click on the “Users” link on the left.
- Create a new user.
- Fill in your user names, ensure the “create access keys” checkbox is checked.
- Click Create.
- Download the credentials. You will use these later.
- Click Close.
- Click on the new user.
- Click on Attach User Policy.
- Click on the radio button next to Custom Policy.
- Click on Select
- Enter a name, which only needs to make sense to you. It won’t be used elsewhere.
- Paste the following policy and replace “my bucket” with the name of the bucket you created.
Configuring the cdn-asset-pipeline Plugin
Add the plugin to your Grails application:
An example configuration is provided below. Note that this is only an example, it is possible to publish to multiple CDNs. See the cdn-asset-pipeline documentation for details.
Amazon S3 URL
The storage path and therefore the S3 URL can be configured in Config.groovy with Groovy. There are two configurations that are affected by the storage path: the upload configuration and the assets URL. See “grails.assets.cdn.storagePath” and “grails.assets.url” above. The recommended approach is to include the application version.
Amazon S3 allows each resource to provide cache control headers. This is configured with the “expires” key shown above. A long expiry time can be set because the assets-pipeline plugin creates files with a hash based on the content.
Credentials are configured using the “accessKey” and “secretKey” properties. For security purposes, it is recommended to maintain credentials in environment variables. Never commit credentials to a public repository. In this case, the credential user name is kept in CDN_AWS_ACCESS_KEY_ID and the secret in CDN_AWS_SECRET_ACCESS_KEY.
Trimming Uploaded Assets
All assets produced by the asset-pipeline plugin are included in the CDN upload. The number of assets can be trimmed by excluding partial assets, i.e. assets that aren’t directly accessed by the application. An example follows. It assumes assets such as angularjs and bootstrap are included in other files, such as application.js and application.css. Make sure to tailor this to your configuration or your application won’t find the files.
Pushing Content to CDN
$ grails asset-cdn-push
That’s it! The plugin will build assets and upload them. It is smart enough to only upload files not already present in the CDN. Make sure to add this command to your CI/CD environment. Try it by running your application in production mode:
$ grails prod run-war
Using a CDN is a best practice to improve the user experience of your application. For Grails applications, the combination of the asset-pipeline and cdn-asset-pipeline plugins make it easy to maintain your static assets on one or more CDNs.
One thought on “Grails cdn-asset-pipeline Plugin for Improving Web Site Performance”
Brandon Fish pointed out that S3 isn’t a CDN, but one can configure Cloud Front to point to S3. Update the asset URL with the Cloud Front URL and you’re good to go.
If application is run with command ‘grails prod run-app’, then there will be problem not getting resources. Instead of that, application should be run with command ‘grails prod run-war’ or create war and deployed in tomcat.
This question in stackoverflow has problem not getting resource http://stackoverflow.com/questions/29586459/assets-are-fetched-without-the-hash-code-appended-while-using-grails-cdn-asset-p
Thanks! The post has been updated.