Files
ModelForge/backend/alembic/versions/20260825_0004_m2_model_registry.py
T

171 lines
7.6 KiB
Python

"""M2 operational model registry and artifact provenance.
Revision ID: 20260825_0004
Revises: 20260825_0003
"""
import sqlalchemy as sa
from alembic import op
revision = "20260825_0004"
down_revision = "20260825_0003"
branch_labels = None
depends_on = None
def upgrade() -> None:
for column in (
sa.Column("source_type", sa.String(32), nullable=False, server_default="huggingface"),
sa.Column("upstream_metadata", sa.JSON(), nullable=False, server_default=sa.text("'{}'")),
sa.Column("local_metadata", sa.JSON(), nullable=False, server_default=sa.text("'{}'")),
sa.Column(
"interpretation_metadata", sa.JSON(), nullable=False, server_default=sa.text("'{}'")
),
sa.Column("description", sa.Text()),
):
op.add_column("models", column)
op.add_column("model_revisions", sa.Column("archived_at", sa.DateTime(timezone=True)))
for column in (
sa.Column("status", sa.String(32), nullable=False, server_default="remote"),
sa.Column(
"verification_details", sa.JSON(), nullable=False, server_default=sa.text("'{}'")
),
sa.Column("archived_at", sa.DateTime(timezone=True)),
):
op.add_column("model_artifacts", column)
op.alter_column("model_artifacts", "storage_uri", existing_type=sa.Text(), nullable=True)
for column in (
sa.Column(
"transformation_type", sa.String(64), nullable=False, server_default="conversion"
),
sa.Column(
"environment_snapshot", sa.JSON(), nullable=False, server_default=sa.text("'{}'")
),
sa.Column("status", sa.String(32), nullable=False, server_default="remote"),
sa.Column(
"verification_details", sa.JSON(), nullable=False, server_default=sa.text("'{}'")
),
):
op.add_column("derived_artifacts", column)
op.alter_column("derived_artifacts", "storage_uri", existing_type=sa.Text(), nullable=True)
op.alter_column(
"derived_artifacts", "source_artifact_id", existing_type=sa.Uuid(), nullable=True
)
op.create_table(
"storage_roots",
sa.Column("compute_node_id", sa.Uuid(), nullable=False),
sa.Column("name", sa.String(128), nullable=False),
sa.Column("purpose", sa.String(64), nullable=False, server_default="model_artifacts"),
sa.Column("path", sa.Text(), nullable=False),
sa.Column("status", sa.String(32), nullable=False, server_default="unknown"),
sa.Column("writable", sa.Boolean(), nullable=False, server_default=sa.false()),
sa.Column("capacity_bytes", sa.BigInteger()),
sa.Column("free_bytes", sa.BigInteger()),
sa.Column("reserve_bytes", sa.BigInteger(), nullable=False, server_default="0"),
sa.Column("reserve_percent", sa.Integer(), nullable=False, server_default="10"),
sa.Column("capacity_observed_at", sa.DateTime(timezone=True)),
sa.Column("validation_details", sa.JSON(), nullable=False, server_default=sa.text("'{}'")),
sa.Column("deprecated_at", sa.DateTime(timezone=True)),
sa.Column("id", sa.Uuid(), primary_key=True),
sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
sa.ForeignKeyConstraint(["compute_node_id"], ["compute_nodes.id"], ondelete="RESTRICT"),
sa.UniqueConstraint("compute_node_id", "path", name="uq_storage_root_node_path"),
sa.CheckConstraint("reserve_bytes >= 0", name="ck_storage_root_reserve_bytes"),
sa.CheckConstraint(
"reserve_percent >= 0 AND reserve_percent <= 100",
name="ck_storage_root_reserve_percent",
),
)
op.create_index("ix_storage_roots_compute_node_id", "storage_roots", ["compute_node_id"])
op.create_table(
"derived_artifact_sources",
sa.Column("derived_artifact_id", sa.Uuid(), nullable=False),
sa.Column("source_artifact_id", sa.Uuid(), nullable=False),
sa.Column("ordinal", sa.Integer(), nullable=False, server_default="0"),
sa.Column("source_sha256", sa.String(64), nullable=False),
sa.ForeignKeyConstraint(
["derived_artifact_id"], ["derived_artifacts.id"], ondelete="RESTRICT"
),
sa.ForeignKeyConstraint(
["source_artifact_id"], ["model_artifacts.id"], ondelete="RESTRICT"
),
sa.PrimaryKeyConstraint("derived_artifact_id", "source_artifact_id"),
sa.UniqueConstraint(
"derived_artifact_id", "source_artifact_id", name="uq_derived_artifact_source"
),
)
op.create_table(
"artifact_locations",
sa.Column("artifact_id", sa.Uuid()),
sa.Column("derived_artifact_id", sa.Uuid()),
sa.Column("storage_root_id", sa.Uuid(), nullable=False),
sa.Column("relative_path", sa.Text(), nullable=False),
sa.Column("status", sa.String(32), nullable=False, server_default="unknown"),
sa.Column("size_bytes", sa.BigInteger()),
sa.Column("observed_sha256", sa.String(64)),
sa.Column("last_checked_at", sa.DateTime(timezone=True)),
sa.Column("id", sa.Uuid(), primary_key=True),
sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
sa.ForeignKeyConstraint(["artifact_id"], ["model_artifacts.id"], ondelete="RESTRICT"),
sa.ForeignKeyConstraint(
["derived_artifact_id"], ["derived_artifacts.id"], ondelete="RESTRICT"
),
sa.ForeignKeyConstraint(["storage_root_id"], ["storage_roots.id"], ondelete="RESTRICT"),
sa.CheckConstraint(
"(artifact_id IS NOT NULL AND derived_artifact_id IS NULL) OR "
"(artifact_id IS NULL AND derived_artifact_id IS NOT NULL)",
name="ck_artifact_location_one_owner",
),
sa.UniqueConstraint(
"storage_root_id", "relative_path", name="uq_storage_root_relative_path"
),
)
op.create_index("ix_artifact_locations_artifact_id", "artifact_locations", ["artifact_id"])
op.create_index(
"ix_artifact_locations_derived_artifact_id",
"artifact_locations",
["derived_artifact_id"],
)
op.create_index(
"ix_artifact_locations_storage_root_id", "artifact_locations", ["storage_root_id"]
)
def downgrade() -> None:
op.drop_table("artifact_locations")
op.drop_table("derived_artifact_sources")
op.drop_table("storage_roots")
op.execute(
"UPDATE model_artifacts SET storage_uri = 'registry://artifact/' || id::text "
"WHERE storage_uri IS NULL"
)
op.execute(
"UPDATE derived_artifacts SET storage_uri = 'registry://derived/' || id::text "
"WHERE storage_uri IS NULL"
)
op.alter_column("derived_artifacts", "source_artifact_id", existing_type=sa.Uuid(), nullable=False)
op.alter_column("derived_artifacts", "storage_uri", existing_type=sa.Text(), nullable=False)
for column in (
"verification_details",
"status",
"environment_snapshot",
"transformation_type",
):
op.drop_column("derived_artifacts", column)
op.alter_column("model_artifacts", "storage_uri", existing_type=sa.Text(), nullable=False)
for column in ("archived_at", "verification_details", "status"):
op.drop_column("model_artifacts", column)
op.drop_column("model_revisions", "archived_at")
for column in (
"description",
"interpretation_metadata",
"local_metadata",
"upstream_metadata",
"source_type",
):
op.drop_column("models", column)