UNMAINTAINED
This git repository is now unmaintained by Gilt. Development of sbt-codedeploy continues at https://github.com/myyk/sbt-codedeploy, please submit bug reports and feature requests there. Thank you!
SBT plugin for AWS CodeDeploy
Creates zip files for usage with AWS CodeDeploy. Uses defaults that support conventions set forth in the AWS CodeDeploy docs.
WARNING
This plugin intereacts with AWS resources. You will be billed for the AWS resources used by using this plugin.
Installation
Add the following to your project/plugins.sbt file:
addSbtPlugin("com.gilt" % "sbt-codedeploy" % "0.4.0")
Add to your build.sbt or Build.scala:
enablePlugins(CodeDeployPlugin)
SBT CodeDeploy uses the AWS CodeDeploy API to upload the zip to a S3 Bucket (single region-only). You must specify the bucket in your build.sbt or Build.scala:
codedeployBucket in ThisBuild := "your-codedeploy-bucket-name-here"
AWS Setup
You'll also need to have AWS Credentials setup with enough privileges for the sbt-codedeploy to modify your AWS resources. Some great resources for this are::
Conventions
sbt-codedeploy makes use of defaults conventions to make configuring your application easy. Though we find them useful, you may override them to fit your needs.
A simple project might have structure like this in addition to it's regular project files.
/src/codedeploy/content/
/src/codedeploy/scripts/ApplicationStop/
/src/codedeploy/scripts/BeforeInstall/
/src/codedeploy/scripts/AfterInstall/
/src/codedeploy/scripts/ApplicationStart/
/src/codedeploy/scripts/ValidateService/
Source Directory
This is the folder where 'content' and scripts should exist.
Default: /src/codedeploy/
Override: sourceDirectory in CodeDeploy
AppSpec File location
It's assumed that the appspec will be generated by the plugin unless a location is specified. The file will be placed in the root directory of the application's source content's directory structure. By default this is appspec.yml.
Default: None
Override: codedeployAppSpecLocation
Content
The content will be put into the zip at content/.
Defaults: {sourceDirectory in CodeDeploy}/content
Scripts
The scripts will be put into the zip at scripts/.
The plugin is expecting scripts to be located in directories named after the hook they will be used with. If generating the appspec.yml the scripts will be run in an alphabetical ordering. We suggest putting numbers before the script name for easy ordering such as 200KillZombieChildren.sh and 666InstallDaemons.sh which would run in that order.
scripts/ApplicationStop/
scripts/BeforeInstall/
scripts/AfterInstall/
scripts/ApplicationStart/
scripts/ValidateService/
Defaults: {sourceDirectory in CodeDeploy}/scripts
Application Name
The default value is the project's name.
Override like this:
name in CodeDeploy := "new-better-shinier-name"
AWS Profile
The AWS profile that will be used to when making AWS API calls.
Default: "default"
Override: codedeployAwsProfile
Deployment IgnoreApplicationStopFailures
This is a setting used during deployments. It's documented in AWS's CodeDeploy docs here: [DeploymentInfo IgnoreApplicationStopFailures(http://docs.aws.amazon.com/codedeploy/latest/APIReference/API_DeploymentInfo.html#-Type-DeploymentInfo-IgnoreApplicationStopFailures)]
Default: false
Override: codedeployIgnoreApplicationStopFailures
DeploymentGroups
There is an sbt configuration tied to each DeploymentGroup. staging and production are provided. If you need anything else you just add this to your build.sbt. You can add as many DeploymentGroups as you need.
lazy val DeploymentGroupName = config("deploymentGroupName")
GiltCodedeployPlugin.makeCodeDeployConfig(DeploymentGroupName)
If these configs are common with the sbt-cloudformation configs, they can be used to create DeploymentGroups through the plugin. In the CloudFormation stack associated with the config, the outputs "AutoScalingGroupArn" and "CodeDeployTrustRoleArm" must be set.
Usage
Push
If you just want to create the zip and push it to your configured S3 bucket use this.
sbt codedeployPush
Create Deployment
To create a deployment the codedeployCreateDeployment task needs the name of the CodeDeploy DeploymentGroup to deploy to. You must call codedeployPush first to make sure the application revision is available.
sbt “staging:codedeployCreateDeployment”
Or
sbt “production:codedeployCreateDeployment”
You may want to create a new alias such as deploy-staging to clean up the verbose syntax for your specific use case.
Note: codedeployCreateDeployment will create an Application if one does not already exist for the project. It will also try to create a DeploymentGroup if one doesn't already exist with the same name as the config, such as production.
This call does not block for the deployment to finish.
Advanced Overriding
AWS Region
By default, the US East 1 region is used as it's the default in the SDK.
To override in build.sbt or Build.scala:
codedeployRegion := com.amazonaws.regions.Regions.US_WEST_2
Native Packager
See sbt-native-packager-example for an example of how to use with sbt-native-packager.
Content Mappings
If the default content mappings above are not enough you can override them. For example, maybe there is content in multiple places of the classpath that you would like included instead of the one place specified. You can override: codedeployContentMappings.
Script Mappings
If the default script mappings above are not enough you can override them. For example, maybe there are scripts in multiple places of the classpath that you would like included instead of the one place specified. You can override: codedeployScriptMappings.
FAQ
I'm getting an error like this:
[error] (*:codedeployPush) com.amazonaws.services.codedeploy.model.ApplicationDoesNotExistException: Applications not found for xxxxxxxxxxxx (Service: AmazonCodeDeploy; Status Code: 400; Error Code: ApplicationDoesNotExistException; Request ID: xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx)
Check that the name matches your CodeDeploy application name sbt inspect CodeDeploy:name.
If that's fine then, maybe the AWS region is not correct. Follow the steps above to override the default.
My deployment keeps failing.
Make sure you've got an zip of the application revision in your S3 bucket.
My revision is in S3 but my deployment keeps failing.
There is a bug as of now (2015-Mar) with using multi-region S3 buckets. You can only use a bucket that's only in the instance that your EC2 instances are in currently.