Architecture

awspub has some architectural choices which are important to know:

• image names need to be unique within a used account. awspub does detect if an image already exists by querying available images by name. If the name doesn’t exist, it will create the image. If the name exists, it will use that image. If the name exist multiple times it will throw an exception.

• images will not be modified. so if a configuration changes its parameters for an image and the image already exists, the parameters will not be changed on EC2 (also most parameters can’t be changed anyway for an image on EC2).

• snapshots are tracked by a sha256sum of the underlying source file (usually a .vmdk file). Some configration parameters (separate_snapshot and billing_products) do adjust that sha256sum to make it unique for the combination of source .vmdk file and config options.

• only EBS (no instance-store) and HVM (no PV) are supported.

• S3 uploads are using a multipart upload so interrupted uploads can be retried but it also means that interrupted updates need to be cleanup up (best via a bucket lifecycle config