Replace undefined confidence with unknown severity for vulnerabilities

Add index for large set of data
Add test for testing data

db/post_migrate/20200511092714_update_undefined_confidence_from_vulnerabilities.rb

+# frozen_string_literal: true
+
+class UpdateUndefinedConfidenceFromVulnerabilities < ActiveRecord::Migration[6.0]
+  include Gitlab::Database::MigrationHelpers
+
+  INDEX_NAME = 'index_vulnerability_on_id_and_confidence_eq_zero'
+  DOWNTIME = false
+
+  disable_ddl_transaction!
+  BATCH_SIZE = 1_000
+  INTERVAL = 2.minutes
+
+  # 87_602 records to be updated on GitLab.com
+  def up
+    # create temporary index for undefined vulnerabilities
+    add_concurrent_index(:vulnerabilities, :id, where: 'confidence = 0', name: INDEX_NAME)
+
+    return unless Gitlab.ee?
+
+    migration = Gitlab::BackgroundMigration::RemoveUndefinedVulnerabilityConfidenceLevel
+    migration_name = migration.to_s.demodulize
+    relation = migration::Vulnerability.undefined_confidence
+    queue_background_migration_jobs_by_range_at_intervals(relation,
+                                                          migration_name,
+                                                          INTERVAL,
+                                                          batch_size: BATCH_SIZE)
+  end
+
+  def down
+    # no-op
+    # temporary index is to be dropped in a different migration in an upcoming release
+    remove_concurrent_index(:vulnerabilities, :id, where: 'confidence = 0', name: INDEX_NAME)
+    # This migration can not be reversed because we can not know which records had undefined confidence
+  end
+end

db/structure.sql

@@ -10843,6 +10843,8 @@ CREATE UNIQUE INDEX index_vulnerability_occurrences_on_uuid ON public.vulnerabil
 
 CREATE INDEX index_vulnerability_occurrences_on_vulnerability_id ON public.vulnerability_occurrences USING btree (vulnerability_id);
 
+CREATE INDEX index_vulnerability_on_id_and_confidence_eq_zero ON public.vulnerabilities USING btree (id) WHERE (confidence = 0);
+
 CREATE UNIQUE INDEX index_vulnerability_scanners_on_project_id_and_external_id ON public.vulnerability_scanners USING btree (project_id, external_id);
 
 CREATE UNIQUE INDEX index_vulnerability_user_mentions_on_note_id ON public.vulnerability_user_mentions USING btree (note_id) WHERE (note_id IS NOT NULL);
@@ -13764,6 +13766,7 @@ COPY "schema_migrations" (version) FROM STDIN;
 20200506154421
 20200507221434
 20200508091106
+20200511092714
 20200511145545
 \.
 

ee/changelogs/unreleased/remove-undefined-confidence-vulnerability.yml

+---
+title: Replace undefined confidence with unknown severity for vulnerabilities
+merge_request: 31593
+author:
+type: other

ee/lib/ee/gitlab/background_migration/remove_undefined_vulnerability_confidence_level.rb

+# frozen_string_literal: true
+
+module EE
+  module Gitlab
+    module BackgroundMigration
+      module RemoveUndefinedVulnerabilityConfidenceLevel
+        extend ::Gitlab::Utils::Override
+
+        class Vulnerability < ActiveRecord::Base
+          include ::EachBatch
+
+          self.table_name = 'vulnerabilities'
+
+          CONFIDENCE_LEVELS = {
+            undefined: 0,
+            unknown: 2
+          }.with_indifferent_access.freeze
+
+          enum confidence: CONFIDENCE_LEVELS
+
+          def self.undefined_confidence
+            where(confidence: Vulnerability.confidences[:undefined])
+          end
+        end
+
+        override :perform
+        def perform(start_id, stop_id)
+          Vulnerability.undefined_confidence
+                    .where(id: start_id..stop_id)
+                    .update_all(confidence: Vulnerability.confidences[:unknown])
+        end
+      end
+    end
+  end
+end

ee/spec/lib/ee/gitlab/background_migration/remove_undefined_vulnerability_confidence_level_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Gitlab::BackgroundMigration::RemoveUndefinedVulnerabilityConfidenceLevel, :migration, schema: 20200511092714 do
+  let(:vulnerabilities) { table(:vulnerabilities) }
+  let(:identifiers) { table(:vulnerability_identifiers) }
+  let(:projects) { table(:projects) }
+  let(:users) { table(:users) }
+
+  it 'updates undefined confidence level to unknown' do
+    projects.create!(id: 123, namespace_id: 12, name: 'gitlab', path: 'gitlab')
+    users.create!(id: 13, email: 'author@example.com', notification_email: 'author@example.com', name: 'author', username: 'author', projects_limit: 10, state: 'active')
+
+    vul1 = vulnerabilities.create!(vuln_params)
+    vulnerabilities.create!(vuln_params)
+    vul3 = vulnerabilities.create!(vuln_params.merge(confidence: 2))
+
+    expect(vulnerabilities.where(confidence: 2).count). to eq(1)
+    expect(vulnerabilities.where(severity: 5).count). to eq(3)
+
+    described_class.new.perform(vul1.id, vul3.id)
+
+    expect(vulnerabilities.where(confidence: 2).count).to eq(3)
+  end
+
+  def vuln_params
+    {
+      title: 'title',
+      state: 1,
+      severity: 5,
+      confidence: 0,
+      report_type: 2,
+      project_id: 123,
+      author_id: 13
+    }
+  end
+end

ee/spec/migrations/update_undefined_confidence_from_vulnerabilities_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+require Rails.root.join('db', 'post_migrate', '20200511092714_update_undefined_confidence_from_vulnerabilities.rb')
+
+describe UpdateUndefinedConfidenceFromVulnerabilities, :migration do
+  let(:vulnerabilities) { table(:vulnerabilities) }
+  let(:identifiers) { table(:vulnerability_identifiers) }
+  let(:projects) { table(:projects) }
+  let(:users) { table(:users) }
+
+  before do
+    projects.create!(id: 123, namespace_id: 12, name: 'gitlab', path: 'gitlab')
+    users.create!(id: 13, email: 'author@example.com', notification_email: 'author@example.com', name: 'author', username: 'author', projects_limit: 10, state: 'active')
+    stub_const("#{described_class}::BATCH_SIZE", 2)
+  end
+
+  it 'updates undefined confidence levels to unkown', :sidekiq_might_not_need_inline do
+    allow_any_instance_of(Gitlab).to receive(:ee?).and_return(true)
+
+    vulnerabilities.create!(vuln_params)
+    vulnerabilities.create!(vuln_params.merge(confidence: 2))
+
+    expect(vulnerabilities.where(confidence: 0).count). to eq(1)
+
+    migrate!
+
+    expect(vulnerabilities.exists?(confidence: 0)).to be_falsy
+    expect(vulnerabilities.where(confidence: 2).count).to eq(2)
+  end
+
+  it 'skips migration for ce' do
+    allow_any_instance_of(Gitlab).to receive(:ee?).and_return(false)
+
+    vulnerabilities.create!(vuln_params)
+
+    expect(vulnerabilities.where(confidence: 0).count). to eq(1)
+
+    migrate!
+
+    expect(vulnerabilities.exists?(confidence: 0)).to be_truthy
+  end
+
+  def vuln_params
+    {
+      title: 'title',
+      state: 1,
+      confidence: 0,
+      severity: 5,
+      report_type: 2,
+      project_id: 123,
+      author_id: 13
+    }
+  end
+end

lib/gitlab/background_migration/remove_undefined_vulnerability_confidence_level.rb

+# frozen_string_literal: true
+# rubocop:disable Style/Documentation
+
+module Gitlab
+  module BackgroundMigration
+    class RemoveUndefinedVulnerabilityConfidenceLevel
+      def perform(start_id, stop_id)
+      end
+    end
+  end
+end
+
+Gitlab::BackgroundMigration::RemoveUndefinedVulnerabilityConfidenceLevel.prepend_if_ee('EE::Gitlab::BackgroundMigration::RemoveUndefinedVulnerabilityConfidenceLevel')