## Upgrade test scenarios

### File plugin

#### Resources present before upgrade

1. FileRepository with a single version synced from a FileRemote.
2. Half artifacts synced, half not? (test remote artifacts) (dalley)
3. FileRemote with a URL pointing to a repository with content set that's different from the repository version in 1.
4. FilePublication that is created from the repository version in 1.
5. FileDistribution A that points to the publication in 3.
6. FileRepository
7. Test downloading downloaded + lazy content from repository

### RPM plugin

### Container plugin

### Ansible plugin

### Maven plugin

### Brainstorming
* from which version to which version?
* from the latest x.y.z release of pulpcore and plugin being to tested to the master branch of pulpcore and master branch of plugin
* which database tables should we test?
* all the content types, all remote types, all exporters, ALL resources
* each plugin team should decide it
* we want to test:
* migrations
* REST API
* have each plugin upgrade test be a separate job
* stretch goal - test all katello plugins together
* add a check to regular CI that ensures no new migrations are generated when makemigrations is run
* https://pulp.plan.io/issues/4984
* https://pulp.plan.io/issues/6637 - the epic for improving functional test
* Should we combine 6637 effort with the upgrade testing?
* if we don't combine, we will end up with 2 sets of tests
* need to handle cases where a feature is not present in version N-1 and added in version N
* tests for each plugin are already specific to the version they sit within
* as a first step in this giant effort we should simply test that we can install a previous version of pulpcore + plugin and then run migrations for master branch of pulpcore + plugin.

# Ideas from June 24

* Requirements:
* Create a set of tests that are run before an upgrade for pulp_file
* Create a set of unique fixtures for pulp_file tests involved during upgrade testing
* Remove the use of orphan cleanup for tests that are used by post_upgrade tests
* Create a set of post-upgrade tests that rely on data generated by pre-upgrade tests
*