give you indication of which resource failed to update • Delete offending resource or its dependencies manually • Run delete stack again • When in real shit, call support
• http://docs.aws.amazon.com/AWSCloudFormation/latest/ UserGuide/cfn-using-cloudformer.html • http://aws.amazon.com/developertools/6460180344805680 • Uses a Micro instance inside your VPC to interrogate and dumps results to S3 • Parse with the ruby cloudformation-ruby-dsl :-)
Use a phased approach. Use Ref to refer to existing resources, passed as parameters. • Usually in reverse of your pipeline — machines first, VPCs last • e.g. "VpcId" : { "Ref" : "VpcId" }
• Use UpdatePolicy for ASG changes. • Use DeletionPolicy to ensure are not deleted by CloudFormation (instance termination protection on steroids) • Note the Update requires property in the documentation
production. Smaller resource types can help reduce costs. • Parameterise everything. • Use the http://docs.aws.amazon.com/AWSCloudFormation/ latest/UserGuide/template-reference.html