Add repository and activity timestamp indexes for issues and pull requests

collectoss/application/db/models/data.py

@@ -1439,6 +1439,8 @@ class Issue(Base):
         Index("issues_ibfk_1", "repo_id"),
         Index("issues_ibfk_2", "reporter_id"),
         Index("issues_ibfk_4", "pull_request_id"),
+        Index("issues_idx_repo_id_updated_at", "repo_id", "updated_at"),
+        Index("issues_idx_repo_id_created_at", "repo_id", "created_at"),
 
         UniqueConstraint("repo_id", "gh_issue_id"),
         UniqueConstraint("issue_url", name="issue-insert-unique"),
@@ -1747,6 +1749,8 @@ class PullRequest(Base):
         Index(
             "pull_requests_idx_repo_id_data_datex", "repo_id", "data_collection_date"
         ),
+        Index("pull_requests_idx_repo_id_pr_updated_at", "repo_id", "pr_updated_at"),
+        Index("pull_requests_idx_repo_id_pr_created_at", "repo_id", "pr_created_at"),
         Index("pr_ID_prs_table", "pull_request_id"),
         {"schema": "data"},
     )

collectoss/application/schema/alembic/versions/45_add_issue_pr_activity_indexes.py

@@ -0,0 +1,56 @@
+"""add issue and pull request activity indexes
+
+Revision ID: 45
+Revises: 44
+
+"""
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = '45'
+down_revision = '44'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # Concurrent builds keep these tables writable during collection.
+    with op.get_context().autocommit_block():
+        op.create_index(
+            'issues_idx_repo_id_updated_at', 'issues', ['repo_id', 'updated_at'],
+            schema='data', postgresql_concurrently=True,
+        )
+        op.create_index(
+            'issues_idx_repo_id_created_at', 'issues', ['repo_id', 'created_at'],
+            schema='data', postgresql_concurrently=True,
+        )
+        op.create_index(
+            'pull_requests_idx_repo_id_pr_updated_at', 'pull_requests',
+            ['repo_id', 'pr_updated_at'],
+            schema='data', postgresql_concurrently=True,
+        )
+        op.create_index(
+            'pull_requests_idx_repo_id_pr_created_at', 'pull_requests',
+            ['repo_id', 'pr_created_at'],
+            schema='data', postgresql_concurrently=True,
+        )
+
+
+def downgrade():
+    with op.get_context().autocommit_block():
+        op.drop_index(
+            'pull_requests_idx_repo_id_pr_created_at', table_name='pull_requests',
+            schema='data', postgresql_concurrently=True,
+        )
+        op.drop_index(
+            'pull_requests_idx_repo_id_pr_updated_at', table_name='pull_requests',
+            schema='data', postgresql_concurrently=True,
+        )
+        op.drop_index(
+            'issues_idx_repo_id_created_at', table_name='issues',
+            schema='data', postgresql_concurrently=True,
+        )
+        op.drop_index(
+            'issues_idx_repo_id_updated_at', table_name='issues',
+            schema='data', postgresql_concurrently=True,
+        )