UCD should impose an order on the modules of the user-data file

[markdryan]

The top level of a user-data file is a YAML mapping. YAML mappings are not ordered. Despite this, UCD processes the modules in the user data file in the order they appear in the file. This is problematic for a number of reasons.

1. It's just luck, or rather a side effect of the YAML parser used, that it works. The YAML document defines no explicit ordering.

2. Other implementations of cloud-init do not do this. As the YAML document does not contain any ordering information, they rely on a separate configuration file, e.g., /etc/cloud/cloud.cfg, to determine which modules to run and in what order. The implications of this is that you cannot safely use an Ubuntu or a Fedora user data file on Clearlinux as the chances are, the modules will run in an unexpected order on Clear.

3. It's not possible to use a tool generated cloud-init file with Clearlinux. As a mapping has no ordering, yaml marshalers tend to generate yaml files in which mapping elements are written out in alphabetical order. This means that tools such as ccloudvm, that generate user data files, will run into issues when using UCD in Clearlinux. Users of these tools have no way to control the ordering of modules in the final user data file. This shouldn't be a problem as the ordering should be meaningless. Unfortunately, in UCD the order does have a meaning, so there's an issue.