I'm planning to build some integration testing against a Nomad agent running in dev, all of the scenarios we want to test can be created with job files and shell scripts running inside containers to simulate failures of various kinds.

I'm opening this to discuss the approach these tests should take:

• a shell script using the levant cli
• tests in the levant package (Go)
• a test package using the levant cli (Go)

Test scenarios:

• successful deployment of a simple job
• deployment failure (container exits with code 1)
• driver error (bad docker image tag)
• evaluation failure (unsatisfiable constraint)
• canary auto promote success
• canary auto promote failure

All scenarios can be implemented with the Docker driver and the alpine image.