Smoke unit test for redis migrations

[archibald1418]

Motivation and context

Basic coverage for redis migrations support from #8898

How has this been tested?

Prepare

• A .py migration file with class Migration is generated for each test in the directory of a django application
• The migration file's Migration.run classmethod is patched with unittest.Mock to assert mock_calls from the migration tool to confirm that the migration was discovered and applied
• Redis is patched with fakeredis instance throughout the whole test suite since our main concern is the migration tool itself, not the Redis database.

Case 1: Migration is added and applied successfully

• Generate a correct stub file named 000_first.py with the following contents:

from cvat.apps.redis_handler.migration_loader import BaseMigration

class Migration(BaseMigration):
    @classmethod
    def run(cls): ...

In order for the stub to be discoverable by the migration tool, its name should start with numbers and it should contain a class Migration which is a subclass of BaseMigration.

• Patch run method with a mock.
• Call django's migrateredis.py script with django.management.call_command('migrateredis')
• Assert that engine.001_cleanup_scheduled_jobs and redis_handler.001_first are added as members to Redis set cvat:applied_migrations
• Assert that our run mock was called once

Case 2: Migration is not added and not applied

• Flush fakeredis after previous case
• Generate a bad stub file named 000_second.py with the following contents:

class Migration:
    @classmethod
    def run(cls): ...

Since the Migration class is not a subclass of BaseMigration, we expect that

• LoaderError is raised
• cvat:applied_migrations is an empty set
• Migration.run is not called

Checklist

• I submit my changes into the develop branch
• I have created a changelog fragment
• I have updated the documentation accordingly
• I have added tests to cover my changes
• I have linked related issues (see GitHub docs)

License

• I submit my code changes under the same MIT License that covers the project.
  Feel free to contact the maintainers if that's a concern.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 73.90%. Comparing base (cb7306b) to head (df6a60b).

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9117      +/-   ##
===========================================
+ Coverage    73.83%   73.90%   +0.06%     
===========================================
  Files          431      432       +1     
  Lines        44804    44809       +5     
  Branches      3892     3892              
===========================================
+ Hits         33081    33114      +33     
+ Misses       11723    11695      -28     

Components	Coverage Δ
cvat-ui	77.50% <ø> (-0.03%)	⬇️
cvat-server	70.94% <100.00%> (+0.14%)	⬆️

[Marishka17]

• Could you please explain the purpose of generation test migration? why existing migrations are not enough? I do not really think that we need to modify/add files from cvat/apps when running tests. Otherwise, the generated files may not be removed e.g. when we stop debugging before cleanup is called.
• Do I understand correctly that we need to update this test each time when a new Redis migration is added?

[archibald1418]

All the existing migration files are correct, which means I cannot test a scenario where the tool throws LoaderError.
Though, case 1 could be rewritten to use existing migration files

[archibald1418]

we need to update this test each time when a new Redis migration is added?

I didn't get it, why would that be the case? Tests are using fake Redis database

[Marishka17]

I didn't get it, why would that be the case? Tests are using fake Redis database

I'm talking about the case when we add one more redis migration. In that case the self.assertEqual(conn.smembers("cvat:applied_migrations"), exp_migrations) will fail. Please, check that the expected by the test migration keys are added to cvat:applied_migrations instead of comparing all set members.

[archibald1418]

Yes totally makes sense, applied

[Marishka17]

I guess the more appropriate test should cover a case when migrateredis is called with --check arg.

[archibald1418]

Makes sense

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud