openstack
/
cinder
Code Issues Proposed changes

Remove leftover nested quota DB fields from model 

Nested quotas used a couple of DB fields named allocated and
allocated_id, but nested quotas driver has been gone for a while, and in
W we removed code references to these 2 fields, but we forgot to remove
the reference from the ORM models.

This means that we cannot remove the DB fields yet, as that would break
the rolling upgrades, because SQLAlchemy from the X release would still
try to load the fields based on its ORM models and would break because
it's expecting those 2 fields.  We should have removed the ORM model
references when we removed the code that used those fields.

In this patch we remove the ORM model fields and also leave a note as a
comment with the required changes (migration and tests) for the next
release to complete this process.

This creates an inconsistency between the ORM models in models.py which
would make the test_models_sync test fail.  To prevent it, the patch
also improves the filter_metadata_diff method to support diff directives
in the form of lists and leverages that functionality to be able to
ignore the discrepancies introduced in this patch until we make them be
on sync on the next release with the migration code that is included in
the comments.

Change-Id: I5b89ba78d02c9374a7078607ea4f348a1acc4abd
This commit is contained in:
Gorka Eguileor 2021-10-21 09:46:06 +02:00
parent 402787ffcc
commit 1a9e911ad4
2 changed files with 105 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
cinder/db/sqlalchemy/models.py
View File

@ -804,8 +804,6 @@ class Quota(BASE, CinderBase):
resource = sa.Column(sa.String(300), nullable=False)
hard_limit = sa.Column(sa.Integer, nullable=True)
# TODO: (X release): Remove allocated, belonged to nested quotas
allocated = sa.Column(sa.Integer, default=0)
class QuotaClass(BASE, CinderBase):
@ -880,10 +878,6 @@ class Reservation(BASE, CinderBase):
usage_id = sa.Column(
sa.Integer, sa.ForeignKey('quota_usages.id'), nullable=True, index=True
)
# TODO: (X release): Remove allocated_id, belonged to nested quotas
allocated_id = sa.Column(
sa.Integer, sa.ForeignKey('quotas.id'), nullable=True, index=True
)
project_id = sa.Column(sa.String(255), index=True)
resource = sa.Column(sa.String(300))
@ -898,12 +892,6 @@ class Reservation(BASE, CinderBase):
primaryjoin='and_(Reservation.usage_id == QuotaUsage.id,'
'QuotaUsage.deleted == False)',
)
# TODO: (X release): Remove allocated_id, belonged to nested quotas
quota = relationship(
"Quota",
foreign_keys=allocated_id,
primaryjoin='and_(Reservation.allocated_id == Quota.id)',
)
class Snapshot(BASE, CinderBase):

108
cinder/tests/unit/db/test_migrations.py
View File

@ -112,8 +112,34 @@ class CinderModelsMigrationsSync(test_migrations.ModelsMigrationsSync):
return models.BASE.metadata
def filter_metadata_diff(self, diff):
# Overriding the parent method to decide on certain attributes
# that maybe present in the DB but not in the models.py
"""Filter out allowed differences between DB ORM model and actual DB
We want to keep the DB ORM (models.py) and DB tables in sync, so the
test_models_sync test checks for discrepancies between them.
Due to the rolling upgrades feature there are cases where we will have
the model and DB out of sync for 1 release, as we stop using it in the
ORM first and then remove it from the DB in the next, so we use this
method to allow such discrepancies. We must add a TODO item on the
filtering code with the release the filtering must be removed.
The diff parameter is a list of diff directives, which can be a tuple
or a list of tuples:
https://alembic.sqlalchemy.org/en/latest/api/autogenerate.html#getting-diffs
"""
# TODO: (D Release) Remove this method and its usage
def ignore_leftover_nested_quota(element):
operation = element[0]
if operation == 'remove_column':
table, column = element[2], element[3].name
return (table, column) in (('quotas', 'allocated'),
('reservations', 'allocated_id'))
if operation == 'remove_index':
return element[1].name == 'ix_reservations_allocated_id'
if operation == 'remove_fk':
return (element[1].table.name == 'reservations'
and element[1].column_keys == ['allocated_id'])
return False
def include_element(element):
"""Determine whether diff element should be excluded."""
@ -134,9 +160,31 @@ class CinderModelsMigrationsSync(test_migrations.ModelsMigrationsSync):
('encryption', 'control_location'),
}
if ignore_leftover_nested_quota(element):
return False
return True
return [element for element in diff if include_element(element[0])]
def filter_elements(diff_directive):
"""Return only the elements that should not be ignored.
It may return None or [] when all elements from the directive have
been filtered out.
"""
if isinstance(diff_directive, list):
return [element for element in diff_directive
if include_element(element)]
if include_element(diff_directive):
return diff_directive
return None
result = []
for diff_directive in diff:
remaining = filter_elements(diff_directive)
if remaining:
result.append(remaining)
return result
class TestModelsSyncSQLite(
@ -191,6 +239,10 @@ class MigrationsWalk(
# we added an online migration to set the value, but we also provide
# a default on the OVO, the ORM, and the DB engine.
'9ab1b092a404',
# Removing allocated_id and allocated columns is acceptable now since
# we stopped using them in the code on the previous release.
# TODO: (D Release) Uncomment next line
# 'afd7494d43b7',
]
FORBIDDEN_METHODS = ('alembic.operations.Operations.alter_column',
'alembic.operations.Operations.drop_column',
@ -317,6 +369,56 @@ class MigrationsWalk(
snapshots = db_utils.get_table(connection, 'snapshots')
self.assertFalse(snapshots.c.use_quota.nullable)
# TODO: (D Release) Uncomment method _check_afd7494d43b7 and create a
# migration with hash afd7494d43b7 using the following command:
# $ tox -e venv -- alembic -c cinder/db/alembic.ini revision \
# --rev-id afd7494d43b7 -m 'drop quota leftovers'
# Then replace the upgrade method in file
# cinder/db/migrations/versions/afd7494d43b7_drop_quota_leftovers.py with
# the uncommented upgrade method below, removing the unused sqlalchemy
# import and adding "from oslo_db.sqlalchemy import utils as db_utils"
#
# def _check_afd7494d43b7(self, connection):
# """Test drop allocated related columns."""
# reservations = db_utils.get_table(connection, 'reservations')
# self.assertNotIn('allocated_id', reservations.c)
# quotas = db_utils.get_table(connection, 'quotas')
# self.assertNotIn('allocated', quotas.c)
#
# def upgrade():
# connection = op.get_bind()
# # SQLite doesn't support dropping columns, so we use a workaround
# if connection.engine.name == 'sqlite':
# with op.batch_alter_table('reservations') as batch_op:
# batch_op.drop_index(op.f('ix_reservations_allocated_id'))
# batch_op.drop_column('allocated_id')
# with op.batch_alter_table('quotas') as batch_op:
# batch_op.drop_column('allocated')
#
# else:
# # The foreign key is unnamed and Cinder doesn't set a naming
# # convention, so the name was decided by the DB engine on
# # creation, and each engine uses a different convention, so we
# # find out the name.
# fk_name = db_utils.get_foreign_key_constraint_name(
# connection, 'reservations', 'allocated_id')
# op.drop_constraint(fk_name, 'reservations', type_='foreignkey')
#
# # Find out the name of the index as well
# indexes_names = db_utils.get_indexes(connection, 'reservations')
# index_name = [idx['name']
# for idx in indexes_names
# if idx['column_names'] == ['allocated_id']]
# # There HAS to be an index, but just to be safe...
# if index_name:
# # Use op.f to indicate that the index name already has the
# # naming convention applied to it.
# op.drop_index(op.f(index_name[0]), table_name='reservations')
#
# op.drop_column('reservations', 'allocated_id')
# op.drop_column('quotas', 'allocated')
#
class TestMigrationsWalkSQLite(
MigrationsWalk,