This document contains information on what you need to know about Django Migrations in the Open edX platform. If you are unfamiliar with database migrations, or, specifically, Django migrations, please read the reference documentation. Django does a good job of abstracting away what's behind the scenes when you run "./manage.py makemigrations" and "./manage.py migrate"- understanding what happens during these operations is extremely useful.
If you have non-trivial migrations to apply, or if two non-local environments (e.g. stage and production) have different migration states, describe your situation in the #django Slack channel and go talk to the SRE team before doing anything else. Similarly, migrations can become complicated when two different people create two different migrations around the same time. When in doubt, post in #django and talk to SRE.
It is often useful to review or provide the SQL generated by a migration. See sqlmigrate doc for details.
Django migrations should be considered "applied" as soon as they land on master of a repo. Missing ( ghost ) migrations cause problems for Django, and require manual intervention to fix. Fix forward on migrations. Be sure to properly consider all the points below so that you're less likely to want to delete or change a migration. If you do delete a migration, or revert a commit that contains a migration, follow this guide to communicating with the organization and the community: How to revert a migration from master. Also, you should never roll back migrations (manually) without rolling back the code that relies on them.
If you still think you need to change old migrations, and you want to verify that there isn't an alternative, see the "When in doubt" section.
Also, if you absolutely must change old migrations (this includes things like squashing), after merging and verifying your migration changes, you should update the sql that is used to populate devstack during provisioning. This is to prevent conflicts during provisioning in the future. To update the sql, in the devstack repository, run .update-dbs-init-sql-scripts.sh
. This should update edxapp.sql, edxapp_csmh.sql, and ecommerce.sql. Create a PR with the updated sql and merge to devstack as usual.
Along the same lines of a migration being considered "applied" once merged into master, you should never change the dependencies of a migration once it has landed on master. It will cause real problems and probably downtime for which environment it is deployed to. When you create new migrations in a feature branch, you want those to be the most recent migrations when you merge into master. Using an analogy to git, you always want your new migrations to be at the "HEAD" of the migration history in your app.
Here at edX, we use the blue-green deployment method. The important detail about this deployment method is that, for some period of time, traffic is going to both the old code and new code. That detail is especially important when deploying database migrations that alter database columns and tables in a manner that is not backward-compatible with the previous release.
Let's go through a couple examples with our user table, auth_user. It has a few different columns, but we'll use the full_name column for the examples.
Say we decide to change the column's name from full_name (with an underscore) to fullname (no underscore). Our code in production is using full_name. When it's time to deploy this new release, we simply generate a migration and deploy it. Since we are using blue-green deployments, our old code is still looking for the original column name, full_name. However, the new deployment changed the name to fullname, so the original code starts failing.
Instead of renaming the column, say we delete it completely. Again, the database is modified when we deploy, and the original code that is still running will fail.
Because we operate in an environment where new and old code are running simultaneously against the same database, new code must always be compatible with the older database schema. Newer deployments can add tables and columns, but neither can be deleted unless the old code is no longer referencing the deleted tables or columns.
In the edx-platform codebase there is a unit test test_migrations_are_in_sync
in test_db.py which ensures that django migrations and models are in sync. Migrations to drop columns or tables generally require at least two releases, one which removes references and one which has the drop migration. The first release will fail the unit test. For this reason you will need to skip that unit test during your release sequence and restore it when you are done. This also applies to libraries used by edx-platform, such as edx-proctoring, the test will fail when edx-platform receives the interim version.
The skip should include a ticket number and brief info on what it's for:
@unittest.skip( "Temporary skip for TICKET-1234 while the fnord column is removed from the snood table" ) |
For either a nullable or non-nullable column, first make sure there are no other models or code that actually use the column. If there are, make adjustments to those first before working through the deletion flow.
For NULLABLE columns, this involves TWO releases:
test_migrations_are_in_sync
unit test.test_migrations_are_in_sync
unit test.For NOT-NULLABLE columns, this involves THREE releases:
test_migrations_are_in_sync
unit test.test_migrations_are_in_sync
unit test.Returning to our example with the auth_user table. If we still want to drop the full_name column, we should do the following:
When dropping ManyToManyField columns, consider that the Django ORM uses a complicated automatic mapping to map the field to certain model names. So unlike other columns where it's easy to remove all usages of the column, unexpected column usages can still occur via the Django ORM manager (such as django.db.models.fields.related_descriptors.create_reverse_many_to_one_manager
). So the field removal and the migration should be two separate steps.
So - for dropping ManyToManyField columns, use at least TWO releases:
test_migrations_are_in_sync
unit test. Deploy.test_migrations_are_in_sync
unit test. Deploy.Failing to separate the two steps may result in the field being used by the old code during the blue-green deployment after the migration has been performed, resulting in production errors.
Renaming a column while keeping the business logic fully functional and without taking any down time is a very delicate and complex process. Some things to keep in mind before you start:
THREE releases:
test_migrations_are_in_sync
unit test.makemigrations
, this should pick up the field removal from the previous stage.test_migrations_are_in_sync
unit test.Due to the current workings of the Open edX ecosystem, some 2U-specific required steps are part of this process to avoid unnecessary problems for 2U, and the community as a whole. |
Pre-requirements:
TWO releases (after pre-requirements):
test_migrations_are_in_sync
unit test.Once a table is removed:
app-permissions
PR to remove that group.See Removing a Djangoapp from an existing project.
A good way to think of this is that migrations can "expand" and "contract" the database. Adding fields is an expansion, and removing them is a contraction. If you're feeling a bit more mathematical today, there's a partial ordering relation on (db, code) where your database and code are in the relation iff the set of fields in the DB is a (non-strict) superset of the fields in the code (... well, this isn't quite right, since changing fields is OK in circumstances like extending the length of a CharField
. Defining the relation precisely is left as an exercise to the reader). Under this model, changing a field (say, from a plain CharField
to an EmailField
) would consist of an expansion (adding the EmailField
) followed later (potentially much later, but mainly not in the same release) by a contraction (deleting the CharField
). Code can also expand and contract in similar ways, by changing which fields are declared in your Django models.
For a migration to be backwards compatible, the database must always be at least as "large" as the code. It can be larger (contain a field not referenced by the code), but not smaller.
If you're writing a data migration, don't import the model directly. Instead, allow Django to use the historical version of your model. This will allow your migration step to use the old (historical) version of your model, even if the model will later by changed by a subsequent database migration.
def combine_names(apps, schema_editor): # We can't import the Person model directly as it may be a newer # version than this migration expects. We use the historical version. Person = apps.get_model('yourappname', 'Person') for person in Person.objects.all(): person.name = '%s %s' % (person.first_name, person.last_name) person.save() |
First read "Deployment and backward-compatible migrations" for general information about handling blue-green deployments. Then see this section for special consideration for large or problematic tables.
When using AWS Aurora, a nullable column can be added to existing large (100k rows+?) tables without causing downtime. However, the migration may still timeout in GoCD - so please coordinate the release with SRE.
NOTES:
AWS Aurora Docs: https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/AuroraMySQL.Managing.FastDDL.html
On AWS Aurora, indexes can be build on large tables without causing downtime, but this requires SRE coordination as the migration may timeout in GoCD.
Adding fields to large tables can cause operational issues. What is safe varies from database version (MySQL 5.7 vs. 5.8. vs. 8.0) and specialized backends (like Aurora). Also, even if the database supports adding things in a non-locking way, Django's migrations framework may not understand how to formulate the right SQL to do so.
A lower risk alternative is to create a new model and link it together with a OneToOneField. You can use the primary_key=True
option in order to have the new table's primary key match the values of the parent table.
This does complicate the code somewhat, and you should be careful about avoiding n+1 queries by calling select_related
. The benefit is not having to hold your breath when the migration rolls out, for fear that you just froze a heavily used table and brought down the site.
Top ones (ordered by descending size):
courseware_studentmodulehistory, courseware_studentmodule student_historicalcourseenrollment, student_courseenrollment student_anonymoususerid, user_api_userorgtag django_comment_client_role_users, certificates_generatedcertificate auth_userprofile, user_api_userpreference, auth_user, oauth2_accesstoken |
Top ones (ordered by descending calls-per-minute):
user_api_userpreference auth_user student_understanding theming_sitetheme django_site course_modes_coursemode courseware_studentmodule course_overviews_courseoverview waffle_utils_waffleflagcourseoverridemodel edxval_videoimage edxval_profile completion_blockcompletion student_anonymoususerid |
Migrations are currently not run in unit tests.
The paver commands that kick off the Lettuce and bokchoy tests run migrations. However, because this would take a long time if we started from scratch, we cache the latest state of the database after certain intervals (every couple months when someone checks in a new cache) so all the migrations are not run, but only the ones added since the last time the database state was cached.
Once you are happy with how your fields are defined in models.py, run the following command. The resulting file will be checked in with your PR.
./manage.py [lms|cms] --settings=devstack_docker makemigrations --initial name_of_app |
Making a migration to modify an existing table
When you make changes to your model, create migration file and check it in:
./manage.py [lms|cms] --settings=devstack_docker makemigrations name_of_app --pythonpath=. |
Make sure you are pointing to the correct environment file.
After creating your migration file, if you are running Open edX via the DevStack configuration, it is sometimes useful to review the SQL for your new migration:
./manage.py [lms|cms] --settings=devstack_docker sqlmigrate name_of_app number |
After creating your migration file, if you are running Open edX via the DevStack configuration, you can perform the migration using the following command:
./manage.py [lms|cms] --settings=devstack_docker migrate name_of_app |
./manage.py [lms|cms] --settings=devstack_docker migrate name_of_app <number> |
where <number>
is the prefix of the migration file that you want to roll back to.
There's a special "zero" migration name to unapply all migrations, including the initial migration.
./manage.py [lms|cms] --settings=devstack_docker migrate name_of_app zero |
Example for CSM primary key to bigint migration.
Do the following before merging/deploying the code, otherwise the pipeline will try to run the migrations
# Copy file contents into /edx/app/edxapp/edx-platform/lms/djangoapps/courseware/migrations/0011_csm_id_bigint.py on worker machine # Check that the migration shows up in the list as unapplied root@ip-10-3-71-92:/edx/app/edxapp/edx-platform# /edx/bin/edxapp-migrate-cms --noinput --list courseware sudo: unable to resolve host ip-10-3-71-92 WARNING:py.warnings:/edx/app/edxapp/edx-platform/lms/djangoapps/courseware/__init__.py:7: DeprecationWarning: Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported warnings.warn("Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported", DeprecationWarning) 2019-08-30 18:08:47,084 WARNING 11141 [enterprise.utils] [user None] utils.py:55 - Could not import Registry from third_party_auth.provider 2019-08-30 18:08:47,084 WARNING 11141 [enterprise.utils] [user None] utils.py:56 - cannot import name _LTI_BACKENDS courseware [X] 0001_initial [X] 0002_coursedynamicupgradedeadlineconfiguration_dynamicupgradedeadlineconfiguration [X] 0003_auto_20170825_0935 [X] 0004_auto_20171010_1639 [X] 0005_orgdynamicupgradedeadlineconfiguration [X] 0006_remove_module_id_index [X] 0007_remove_done_index [X] 0008_move_idde_to_edx_when [X] 0009_auto_20190703_1955 [X] 0010_auto_20190709_1559 [ ] 0011_csm_id_bigint sudo: unable to resolve host ip-10-3-71-92 WARNING:py.warnings:/edx/app/edxapp/edx-platform/lms/djangoapps/courseware/__init__.py:7: DeprecationWarning: Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported warnings.warn("Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported", DeprecationWarning) 2019-08-30 18:08:52,683 WARNING 11392 [enterprise.utils] [user None] utils.py:55 - Could not import Registry from third_party_auth.provider 2019-08-30 18:08:52,684 WARNING 11392 [enterprise.utils] [user None] utils.py:56 - cannot import name _LTI_BACKENDS courseware [X] 0001_initial [X] 0002_coursedynamicupgradedeadlineconfiguration_dynamicupgradedeadlineconfiguration [X] 0003_auto_20170825_0935 [X] 0004_auto_20171010_1639 [X] 0005_orgdynamicupgradedeadlineconfiguration [X] 0006_remove_module_id_index [X] 0007_remove_done_index [X] 0008_move_idde_to_edx_when [X] 0009_auto_20190703_1955 [X] 0010_auto_20190709_1559 [ ] 0011_csm_id_bigint root@ip-10-3-71-92:/edx/app/edxapp/edx-platform# |
# Add extra space in front to prevent bash from writing password to history root@ip-10-3-71-92:/edx/app/edxapp/edx-platform# DB_MIGRATION_USER=migrate001 DB_MIGRATION_PASS=redacted /edx/bin/edxapp-migrate-cms --fake courseware 0011_csm_id_bigint # Confirm the migration was applied root@ip-10-3-71-92:/edx/app/edxapp/edx-platform# /edx/bin/edxapp-migrate-cms --list courseware sudo: unable to resolve host ip-10-3-71-92 WARNING:py.warnings:/edx/app/edxapp/edx-platform/lms/djangoapps/courseware/__init__.py:7: DeprecationWarning: Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported warnings.warn("Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported", DeprecationWarning) 2019-08-30 18:55:31,993 WARNING 19468 [enterprise.utils] [user None] utils.py:55 - Could not import Registry from third_party_auth.provider 2019-08-30 18:55:31,993 WARNING 19468 [enterprise.utils] [user None] utils.py:56 - cannot import name EnterpriseCustomerIdentityProvider courseware [X] 0001_initial [X] 0002_coursedynamicupgradedeadlineconfiguration_dynamicupgradedeadlineconfiguration [X] 0003_auto_20170825_0935 [X] 0004_auto_20171010_1639 [X] 0005_orgdynamicupgradedeadlineconfiguration [X] 0006_remove_module_id_index [X] 0007_remove_done_index [X] 0008_move_idde_to_edx_when [X] 0009_auto_20190703_1955 [X] 0010_auto_20190709_1559 [X] 0011_csm_id_bigint sudo: unable to resolve host ip-10-3-71-92 WARNING:py.warnings:/edx/app/edxapp/edx-platform/lms/djangoapps/courseware/__init__.py:7: DeprecationWarning: Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported warnings.warn("Importing 'lms.djangoapps.courseware' as 'courseware' is no longer supported", DeprecationWarning) 2019-08-30 18:55:37,151 WARNING 19583 [enterprise.utils] [user None] utils.py:55 - Could not import Registry from third_party_auth.provider 2019-08-30 18:55:37,152 WARNING 19583 [enterprise.utils] [user None] utils.py:56 - cannot import name EnterpriseCustomerIdentityProvider courseware [X] 0001_initial [X] 0002_coursedynamicupgradedeadlineconfiguration_dynamicupgradedeadlineconfiguration [X] 0003_auto_20170825_0935 [X] 0004_auto_20171010_1639 [X] 0005_orgdynamicupgradedeadlineconfiguration [X] 0006_remove_module_id_index [X] 0007_remove_done_index [X] 0008_move_idde_to_edx_when [X] 0009_auto_20190703_1955 [X] 0010_auto_20190709_1559 [X] 0011_csm_id_bigint |
# Remove the migration file root@ip-10-3-71-92:/edx/app/edxapp/edx-platform# rm /edx/app/edxapp/edx-platform/lms/djangoapps/courseware/migrations/0011_csm_id_bigint.py |
See Django's Documentation for Squashing Migrations. Some useful tips for squashing:
The primary benefit of squashing migrations is the speed-up of running migrations from scratch. If you are not running migrations from scratch, this may not help you.
Pros:
Cons:
Conclusion:
See ARCHBOM-1148 for more details. |
Testing your squashed migrations.
|
This enables you to squash and not mess up systems currently in production that aren’t fully up-to-date yet. The recommended process is to squash, keeping the old files, commit and release, wait until all systems are upgraded with the new release (or if you’re a third-party project, ensure your users upgrade releases in order without skipping any), and then remove the old files, commit and do a second release.