fix(grouping): make id bigint sequence

src/migrations/versions/95b4ba4f731d_migration.py

@@ -0,0 +1,33 @@
+"""Migration
+
+Revision ID: 95b4ba4f731d
+Revises: 2597db647e9a
+Create Date: 2024-07-17 23:35:18.871569
+
+"""
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "95b4ba4f731d"
+down_revision = "2597db647e9a"
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+ op.execute("ALTER TABLE grouping_records ALTER COLUMN id TYPE BIGINT")
+ op.execute("CREATE SEQUENCE IF NOT EXISTS grouping_records_id_seq")
+ op.execute(
+ "ALTER TABLE grouping_records ALTER COLUMN id SET DEFAULT nextval('grouping_records_id_seq')"
+ )
+ op.execute("ALTER SEQUENCE grouping_records_id_seq OWNED BY grouping_records.id")
+ op.execute(
+ "SELECT setval('grouping_records_id_seq', COALESCE((SELECT MAX(id) FROM grouping_records), 1), true)"
+ )
+
+
+def downgrade():
+ op.execute("ALTER TABLE grouping_records ALTER COLUMN id DROP DEFAULT")
+ op.execute("DROP SEQUENCE grouping_records_id_seq")
+ op.execute("ALTER TABLE grouping_records ALTER COLUMN id TYPE INTEGER")

src/seer/db.py

@@ -231,28 +231,61 @@ class DbPrIdToAutofixRunIdMapping(Base):
  )
 
 
-class DbGroupingRecord(Base):
- __tablename__ = "grouping_records"
- id: Mapped[int] = mapped_column(Integer, primary_key=True)
- project_id: Mapped[int] = mapped_column(BigInteger, nullable=False)
+class DbGroupingRecordBase:
+ id: Mapped[int] = mapped_column(
+ BigInteger, primary_key=True, server_default="nextval('grouping_records_id_seq')"
+ )
+ project_id: Mapped[int] = mapped_column(BigInteger, primary_key=True, nullable=False)
  message: Mapped[str] = mapped_column(String, nullable=False)
  error_type: Mapped[str] = mapped_column(String, nullable=True)
  stacktrace_embedding: Mapped[Vector] = mapped_column(Vector(768), nullable=False)
- hash: Mapped[str] = mapped_column(
- String(32), nullable=False, default="00000000000000000000000000000000"
- )
+ hash: Mapped[str] = mapped_column(String(32), nullable=False)
+
 
+class DbGroupingRecord(DbGroupingRecordBase, Base):
+ __tablename__ = "grouping_records"
  __table_args__ = (
  Index(
- "ix_grouping_records_stacktrace_embedding_hnsw",
+ "ix_grouping_records_new_stacktrace_embedding_hnsw",
  "stacktrace_embedding",
  postgresql_using="hnsw",
  postgresql_with={"m": 16, "ef_construction": 200},
  postgresql_ops={"stacktrace_embedding": "vector_cosine_ops"},
  ),
  Index(
- "ix_grouping_records_project_id",
+ "ix_grouping_records_new_project_id",
  "project_id",
  ),
- UniqueConstraint("project_id", "hash", name="u_project_id_hash"),
+ UniqueConstraint("project_id", "hash", name="u_project_id_hash_composite"),
+ {"postgresql_partition_by": "HASH (project_id)"},
  )
+
+
+globals().update(
+ {
+ f"DbGroupingRecord{i}": type(
+ f"DbGroupingRecord{i}",
+ (DbGroupingRecordBase, Base),
+ dict(
+ __tablename__=f"grouping_records_p{i}",
+ __table_args__=(
+ Index(
+ f"grouping_records_new_p{i}_stacktrace_embedding_idx",
+ "stacktrace_embedding",
+ postgresql_using="hnsw",
+ postgresql_with={"m": 16, "ef_construction": 200},
+ postgresql_ops={"stacktrace_embedding": "vector_cosine_ops"},
+ ),
+ Index(
+ f"grouping_records_new_p{i}_project_id_idx",
+ "project_id",
+ ),
+ UniqueConstraint(
+ "project_id", "hash", name=f"grouping_records_new_p{i}_project_id_hash_key"
+ ),
+ ),
+ ),
+ )
+ for i in range(100)
+ }
+)