We have some issue when running acceptance test, more generally
when we use to execute scan cmd multiples times.
We were using global singletons for provider and resources suppliers
managment which lead us to improper state in the second scan run.
We should avoid this in the future and make proper initialization of our
dependencies maybe using a dependency injection container.
- Set all test AZ to us-east-1
- Use terraform overriden env in PreCheck() to ensure mutations are
executed with read write credentials
- Fix hidden error in aws_instance test (tag creation failure was not
handled
- Minor fmt fix
- Add filter on aws_instance test
- Add terraform locks
- Do not run terraform init if already done previously
- Do not use chdir in tests (it could cause race if we want to switch to parallel tests, uses --from arg instead)
- Handle terraform destroy errors
Elie
William Beuil
Elie
Stephane Jourdan
Martin Guibert