Merge branch 'ee-31114-internal-ids-are-not-atomic' into 'master'

EE-port of: Atomic generation of internal ids for issues.

See merge request gitlab-org/gitlab-ee!4937

app/models/concerns/atomic_internal_id.rb

+# Include atomic internal id generation scheme for a model
+#
+# This allows us to atomically generate internal ids that are
+# unique within a given scope.
+#
+# For example, let's generate internal ids for Issue per Project:
+# ```
+# class Issue < ActiveRecord::Base
+#   has_internal_id :iid, scope: :project, init: ->(s) { s.project.issues.maximum(:iid) }
+# end
+# ```
+#
+# This generates unique internal ids per project for newly created issues.
+# The generated internal id is saved in the `iid` attribute of `Issue`.
+#
+# This concern uses InternalId records to facilitate atomicity.
+# In the absence of a record for the given scope, one will be created automatically.
+# In this situation, the `init` block is called to calculate the initial value.
+# In the example above, we calculate the maximum `iid` of all issues
+# within the given project.
+#
+# Note that a model may have more than one internal id associated with possibly
+# different scopes.
+module AtomicInternalId
+  extend ActiveSupport::Concern
+
+  module ClassMethods
+    def has_internal_id(column, scope:, init:) # rubocop:disable Naming/PredicateName
+      before_validation(on: :create) do
+        if read_attribute(column).blank?
+          scope_attrs = { scope => association(scope).reader }
+          usage = self.class.table_name.to_sym
+
+          new_iid = InternalId.generate_next(self, scope_attrs, usage, init)
+          write_attribute(column, new_iid)
+        end
+      end
+
+      validates column, presence: true, numericality: true
+    end
+  end
+
+  def to_param
+    iid.to_s
+  end
+end

app/models/concerns/internal_id.rb → app/models/concerns/nonatomic_internal_id.rb

-module InternalId
+module NonatomicInternalId
   extend ActiveSupport::Concern
 
   included do

app/models/deployment.rb

 class Deployment < ActiveRecord::Base
-  include InternalId
+  include NonatomicInternalId
 
   belongs_to :project, required: true
   belongs_to :environment, required: true

app/models/internal_id.rb

+# An InternalId is a strictly monotone sequence of integers
+# generated for a given scope and usage.
+#
+# For example, issues use their project to scope internal ids:
+# In that sense, scope is "project" and usage is "issues".
+# Generated internal ids for an issue are unique per project.
+#
+# See InternalId#usage enum for available usages.
+#
+# In order to leverage InternalId for other usages, the idea is to
+# * Add `usage` value to enum
+# * (Optionally) add columns to `internal_ids` if needed for scope.
+class InternalId < ActiveRecord::Base
+  belongs_to :project
+
+  enum usage: { issues: 0 }
+
+  validates :usage, presence: true
+
+  REQUIRED_SCHEMA_VERSION = 20180305095250
+
+  # Increments #last_value and saves the record
+  #
+  # The operation locks the record and gathers a `ROW SHARE` lock (in PostgreSQL).
+  # As such, the increment is atomic and safe to be called concurrently.
+  def increment_and_save!
+    lock!
+    self.last_value = (last_value || 0) + 1
+    save!
+    last_value
+  end
+
+  class << self
+    def generate_next(subject, scope, usage, init)
+      # Shortcut if `internal_ids` table is not available (yet)
+      # This can be the case in other (unrelated) migration specs
+      return (init.call(subject) || 0) + 1 unless available?
+
+      InternalIdGenerator.new(subject, scope, usage, init).generate
+    end
+
+    def available?
+      @available_flag ||= ActiveRecord::Migrator.current_version >= REQUIRED_SCHEMA_VERSION # rubocop:disable Gitlab/PredicateMemoization
+    end
+
+    # Flushes cached information about schema
+    def reset_column_information
+      @available_flag = nil
+      super
+    end
+  end
+
+  class InternalIdGenerator
+    # Generate next internal id for a given scope and usage.
+    #
+    # For currently supported usages, see #usage enum.
+    #
+    # The method implements a locking scheme that has the following properties:
+    # 1) Generated sequence of internal ids is unique per (scope and usage)
+    # 2) The method is thread-safe and may be used in concurrent threads/processes.
+    # 3) The generated sequence is gapless.
+    # 4) In the absence of a record in the internal_ids table, one will be created
+    #    and last_value will be calculated on the fly.
+    #
+    # subject: The instance we're generating an internal id for. Gets passed to init if called.
+    # scope: Attributes that define the scope for id generation.
+    # usage: Symbol to define the usage of the internal id, see InternalId.usages
+    # init: Block that gets called to initialize InternalId record if not present
+    #       Make sure to not throw exceptions in the absence of records (if this is expected).
+    attr_reader :subject, :scope, :init, :scope_attrs, :usage
+
+    def initialize(subject, scope, usage, init)
+      @subject = subject
+      @scope = scope
+      @init = init
+      @usage = usage
+
+      raise ArgumentError, 'Scope is not well-defined, need at least one column for scope (given: 0)' if scope.empty?
+
+      unless InternalId.usages.has_key?(usage.to_s)
+        raise ArgumentError, "Usage '#{usage}' is unknown. Supported values are #{InternalId.usages.keys} from InternalId.usages"
+      end
+    end
+
+    # Generates next internal id and returns it
+    def generate
+      subject.transaction do
+        # Create a record in internal_ids if one does not yet exist
+        # and increment its last value
+        #
+        # Note this will acquire a ROW SHARE lock on the InternalId record
+        (lookup || create_record).increment_and_save!
+      end
+    end
+
+    private
+
+    # Retrieve InternalId record for (project, usage) combination, if it exists
+    def lookup
+      InternalId.find_by(**scope, usage: usage_value)
+    end
+
+    def usage_value
+      @usage_value ||= InternalId.usages[usage.to_s]
+    end
+
+    # Create InternalId record for (scope, usage) combination, if it doesn't exist
+    #
+    # We blindly insert without synchronization. If another process
+    # was faster in doing this, we'll realize once we hit the unique key constraint
+    # violation. We can safely roll-back the nested transaction and perform
+    # a lookup instead to retrieve the record.
+    def create_record
+      subject.transaction(requires_new: true) do
+        InternalId.create!(
+          **scope,
+          usage: usage_value,
+          last_value: init.call(subject) || 0
+        )
+      end
+    rescue ActiveRecord::RecordNotUnique
+      lookup
+    end
+  end
+end

app/models/issue.rb

@@ -4,7 +4,7 @@ class Issue < ActiveRecord::Base
   prepend EE::Issue
   prepend EE::RelativePositioning
 
-  include InternalId
+  include AtomicInternalId
   include Issuable
   include Noteable
   include Referable
@@ -28,6 +28,8 @@ class Issue < ActiveRecord::Base
   belongs_to :project
   belongs_to :moved_to, class_name: 'Issue'
 
+  has_internal_id :iid, scope: :project, init: ->(s) { s&.project&.issues&.maximum(:iid) }
+
   has_many :events, as: :target, dependent: :destroy # rubocop:disable Cop/ActiveRecordDependent
 
   has_many :merge_requests_closing_issues,

app/models/merge_request.rb

 class MergeRequest < ActiveRecord::Base
-  include InternalId
+  include NonatomicInternalId
   include Issuable
   include Noteable
   include Referable

app/models/milestone.rb

@@ -8,7 +8,7 @@ class Milestone < ActiveRecord::Base
   Started = MilestoneStruct.new('Started', '#started', -3)
 
   include CacheMarkdownField
-  include InternalId
+  include NonatomicInternalId
   include Sortable
   include Referable
   include StripAttribute

app/models/project.rb

@@ -193,6 +193,8 @@ class Project < ActiveRecord::Base
   has_many :todos
   has_many :notification_settings, as: :source, dependent: :delete_all # rubocop:disable Cop/ActiveRecordDependent
 
+  has_many :internal_ids
+
   has_one :import_data, class_name: 'ProjectImportData', inverse_of: :project, autosave: true
   has_one :project_feature, inverse_of: :project
   has_one :statistics, class_name: 'ProjectStatistics'

changelogs/unreleased/31114-internal-ids-are-not-atomic.yml

+---
+title: Atomic generation of internal ids for issues.
+merge_request: 17580
+author:
+type: other

config/initializers/ar_native_database_types.rb

@@ -4,7 +4,7 @@ module ActiveRecord
   module ConnectionAdapters
     class AbstractMysqlAdapter
       NATIVE_DATABASE_TYPES.merge!(
-        bigserial:  { name: 'bigint(20) auto_increment PRIMARY KEY' }
+        bigserial: { name: 'bigint(20) auto_increment PRIMARY KEY' }
       )
     end
   end

db/migrate/20180305095250_create_internal_ids_table.rb

+class CreateInternalIdsTable < ActiveRecord::Migration
+  include Gitlab::Database::MigrationHelpers
+
+  DOWNTIME = false
+
+  def change
+    create_table :internal_ids, id: :bigserial do |t|
+      t.references :project, null: false, foreign_key: { on_delete: :cascade }
+      t.integer :usage, null: false
+      t.integer :last_value, null: false
+
+      t.index [:usage, :project_id], unique: true
+    end
+  end
+end

db/schema.rb

@@ -1222,6 +1222,14 @@ ActiveRecord::Schema.define(version: 20180309160427) do
 
   add_index "index_statuses", ["project_id"], name: "index_index_statuses_on_project_id", unique: true, using: :btree
 
+  create_table "internal_ids", id: :bigserial, force: :cascade do |t|
+    t.integer "project_id", null: false
+    t.integer "usage", null: false
+    t.integer "last_value", null: false
+  end
+
+  add_index "internal_ids", ["usage", "project_id"], name: "index_internal_ids_on_usage_and_project_id", unique: true, using: :btree
+
   create_table "issue_assignees", id: false, force: :cascade do |t|
     t.integer "user_id", null: false
     t.integer "issue_id", null: false
@@ -2668,6 +2676,7 @@ ActiveRecord::Schema.define(version: 20180309160427) do
   add_foreign_key "gpg_signatures", "projects", on_delete: :cascade
   add_foreign_key "group_custom_attributes", "namespaces", column: "group_id", on_delete: :cascade
   add_foreign_key "index_statuses", "projects", name: "fk_74b2492545", on_delete: :cascade
+  add_foreign_key "internal_ids", "projects", on_delete: :cascade
   add_foreign_key "issue_assignees", "issues", name: "fk_b7d881734a", on_delete: :cascade
   add_foreign_key "issue_assignees", "users", name: "fk_5e0c8d9154", on_delete: :cascade
   add_foreign_key "issue_links", "issues", column: "source_id", name: "fk_c900194ff2", on_delete: :cascade

ee/app/models/ee/epic.rb

@@ -3,7 +3,7 @@ module EE
     extend ActiveSupport::Concern
 
     prepended do
-      include InternalId
+      include NonatomicInternalId
       include Issuable
       include Noteable
       include Referable

ee/spec/models/epic_spec.rb

@@ -20,7 +20,7 @@ describe Epic do
   describe 'modules' do
     subject { described_class }
 
-    it { is_expected.to include_module(InternalId) }
+    it { is_expected.to include_module(NonatomicInternalId) }
   end
 
   describe '.order_start_or_end_date_asc' do

spec/factories/internal_ids.rb

+FactoryBot.define do
+  factory :internal_id do
+    project
+    usage :issues
+    last_value { project.issues.maximum(:iid) || 0 }
+  end
+end

spec/lib/gitlab/import_export/all_models.yml

@@ -313,6 +313,7 @@ project:
 - lfs_file_locks
 - project_badges
 - source_of_merge_requests
+- internal_ids
 award_emoji:
 - awardable
 - user

spec/models/concerns/issuable_spec.rb

@@ -34,7 +34,7 @@ describe Issuable do
     subject { build(:issue) }
 
     before do
-      allow(subject).to receive(:set_iid).and_return(false)
+      allow(InternalId).to receive(:generate_next).and_return(nil)
     end
 
     it { is_expected.to validate_presence_of(:project) }

spec/models/internal_id_spec.rb

+require 'spec_helper'
+
+describe InternalId do
+  let(:project) { create(:project) }
+  let(:usage) { :issues }
+  let(:issue) { build(:issue, project: project) }
+  let(:scope) { { project: project } }
+  let(:init) { ->(s) { s.project.issues.size } }
+
+  context 'validations' do
+    it { is_expected.to validate_presence_of(:usage) }
+  end
+
+  describe '.generate_next' do
+    subject { described_class.generate_next(issue, scope, usage, init) }
+
+    context 'in the absence of a record' do
+      it 'creates a record if not yet present' do
+        expect { subject }.to change { described_class.count }.from(0).to(1)
+      end
+
+      it 'stores record attributes' do
+        subject
+
+        described_class.first.tap do |record|
+          expect(record.project).to eq(project)
+          expect(record.usage).to eq(usage.to_s)
+        end
+      end
+
+      context 'with existing issues' do
+        before do
+          rand(1..10).times { create(:issue, project: project) }
+          described_class.delete_all
+        end
+
+        it 'calculates last_value values automatically' do
+          expect(subject).to eq(project.issues.size + 1)
+        end
+      end
+
+      context 'with concurrent inserts on table' do
+        it 'looks up the record if it was created concurrently' do
+          args = { **scope, usage: described_class.usages[usage.to_s] }
+          record = double
+          expect(described_class).to receive(:find_by).with(args).and_return(nil)    # first call, record not present
+          expect(described_class).to receive(:find_by).with(args).and_return(record) # second call, record was created by another process
+          expect(described_class).to receive(:create!).and_raise(ActiveRecord::RecordNotUnique, 'record not unique')
+          expect(record).to receive(:increment_and_save!)
+
+          subject
+        end
+      end
+    end
+
+    it 'generates a strictly monotone, gapless sequence' do
+      seq = (0..rand(100)).map do
+        described_class.generate_next(issue, scope, usage, init)
+      end
+      normalized = seq.map { |i| i - seq.min }
+
+      expect(normalized).to eq((0..seq.size - 1).to_a)
+    end
+
+    context 'with an insufficient schema version' do
+      before do
+        described_class.reset_column_information
+        expect(ActiveRecord::Migrator).to receive(:current_version).and_return(InternalId::REQUIRED_SCHEMA_VERSION - 1)
+      end
+
+      let(:init) { double('block') }
+
+      it 'calculates next internal ids on the fly' do
+        val = rand(1..100)
+
+        expect(init).to receive(:call).with(issue).and_return(val)
+        expect(subject).to eq(val + 1)
+      end
+    end
+  end
+
+  describe '#increment_and_save!' do
+    let(:id) { create(:internal_id) }
+    subject { id.increment_and_save! }
+
+    it 'returns incremented iid' do
+      value = id.last_value
+
+      expect(subject).to eq(value + 1)
+    end
+
+    it 'saves the record' do
+      subject
+
+      expect(id.changed?).to be_falsey
+    end
+
+    context 'with last_value=nil' do
+      let(:id) { build(:internal_id, last_value: nil) }
+
+      it 'returns 1' do
+        expect(subject).to eq(1)
+      end
+    end
+  end
+end

spec/models/issue_spec.rb

@@ -9,11 +9,17 @@ describe Issue do
   describe 'modules' do
     subject { described_class }
 
-    it { is_expected.to include_module(InternalId) }
     it { is_expected.to include_module(Issuable) }
     it { is_expected.to include_module(Referable) }
     it { is_expected.to include_module(Sortable) }
     it { is_expected.to include_module(Taskable) }
+
+    it_behaves_like 'AtomicInternalId' do
+      let(:internal_id_attribute) { :iid }
+      let(:instance) { build(:issue) }
+      let(:scope_attrs) { { project: instance.project } }
+      let(:usage) { :issues }
+    end
   end
 
   subject { create(:issue) }

spec/models/merge_request_spec.rb

@@ -17,7 +17,7 @@ describe MergeRequest do
   describe 'modules' do
     subject { described_class }
 
-    it { is_expected.to include_module(InternalId) }
+    it { is_expected.to include_module(NonatomicInternalId) }
     it { is_expected.to include_module(Issuable) }
     it { is_expected.to include_module(Referable) }
     it { is_expected.to include_module(Sortable) }

spec/support/shared_examples/models/atomic_internal_id_spec.rb

+require 'spec_helper'
+
+shared_examples_for 'AtomicInternalId' do
+  describe '.has_internal_id' do
+    describe 'Module inclusion' do
+      subject { described_class }
+
+      it { is_expected.to include_module(AtomicInternalId) }
+    end
+
+    describe 'Validation' do
+      subject { instance }
+
+      before do
+        allow(InternalId).to receive(:generate_next).and_return(nil)
+      end
+
+      it { is_expected.to validate_presence_of(internal_id_attribute) }
+      it { is_expected.to validate_numericality_of(internal_id_attribute) }
+    end
+
+    describe 'internal id generation' do
+      subject { instance.save! }
+
+      it 'calls InternalId.generate_next and sets internal id attribute' do
+        iid = rand(1..1000)
+
+        expect(InternalId).to receive(:generate_next).with(instance, scope_attrs, usage, any_args).and_return(iid)
+        subject
+        expect(instance.public_send(internal_id_attribute)).to eq(iid)
+      end
+
+      it 'does not overwrite an existing internal id' do
+        instance.public_send("#{internal_id_attribute}=", 4711)
+
+        expect { subject }.not_to change { instance.public_send(internal_id_attribute) }
+      end
+    end
+  end
+end