Fill project_repositories for hashed storage

This adds a background migration that will ensure all projects that
are on hashed storage have a row in `project_repositories`.

Related issue: https://gitlab.com/gitlab-org/gitlab-ce/issues/48527

4 files changed, 311 insertions, 0 deletions

diff --git a/changelogs/unreleased/tc-backfill-hashed-project_repositories.yml b/changelogs/unreleased/tc-backfill-hashed-project_repositories.yml
new file mode 100644
index 00000000000..90a5c8c4e2c
--- /dev/null
+++ b/changelogs/unreleased/tc-backfill-hashed-project_repositories.yml
@@ -0,0 +1,5 @@
+---
+title: Fill project_repositories for hashed storage projects
+merge_request: 23482
+author:
+type: added
diff --git a/db/post_migrate/20181130102132_backfill_hashed_project_repositories.rb b/db/post_migrate/20181130102132_backfill_hashed_project_repositories.rb
new file mode 100644
index 00000000000..b989d9fb43d
--- /dev/null
+++ b/db/post_migrate/20181130102132_backfill_hashed_project_repositories.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+class BackfillHashedProjectRepositories < ActiveRecord::Migration[5.0]
+  include Gitlab::Database::MigrationHelpers
+
+  DOWNTIME = false
+  BATCH_SIZE     = 1_000
+  DELAY_INTERVAL = 1.minutes
+  MIGRATION      = 'BackfillHashedProjectRepositories'
+
+  disable_ddl_transaction!
+
+  class Project < ActiveRecord::Base
+    include EachBatch
+
+    self.table_name = 'projects'
+  end
+
+  def up
+    queue_background_migration_jobs_by_range_at_intervals(Project, MIGRATION, DELAY_INTERVAL)
+  end
+
+  def down
+    # Since there could have been existing rows before the migration
+    # do not remove anything
+  end
+end
diff --git a/lib/gitlab/background_migration/backfill_hashed_project_repositories.rb b/lib/gitlab/background_migration/backfill_hashed_project_repositories.rb
new file mode 100644
index 00000000000..88696dd1aa6
--- /dev/null
+++ b/lib/gitlab/background_migration/backfill_hashed_project_repositories.rb
@@ -0,0 +1,124 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module BackgroundMigration
+    # Class the will create rows in project_repositories for all
+    # projects that are on hashed storage
+    class BackfillHashedProjectRepositories
+      # Model for a Shard
+      class Shard < ActiveRecord::Base
+        self.table_name = 'shards'
+
+        def self.by_name(name)
+          to_a.detect { |shard| shard.name == name } || create_by(name: name)
+        rescue ActiveRecord::RecordNotUnique
+          retry
+        end
+      end
+
+      # Class that will find or create the shard by name.
+      # There is only a small set of shards, which would not change quickly,
+      # so look them up from memory instead of hitting the DB each time.
+      class ShardFinder
+        def find(name)
+          shards.detect { |shard| shard.name == name } || create!(name)
+        rescue ActiveRecord::RecordNotUnique
+          load!
+          retry
+        end
+
+        private
+
+        def create!(name)
+          Shard.create!(name: name).tap { |shard| @shards << shard }
+        end
+
+        def shards
+          @shards || load!
+        end
+
+        def load!
+          @shards = Shard.all.to_a
+        end
+      end
+
+      # Model for a ProjectRepository
+      class ProjectRepository < ActiveRecord::Base
+        self.table_name = 'project_repositories'
+
+        belongs_to :project, inverse_of: :project_repository
+      end
+
+      # Model for a Project
+      class Project < ActiveRecord::Base
+        self.table_name = 'projects'
+
+        HASHED_PATH_PREFIX = '@hashed'
+        HASHED_STORAGE_FEATURES = {
+          repository: 1,
+          attachments: 2
+        }.freeze
+
+        has_one :project_repository, inverse_of: :project
+
+        class << self
+          def on_hashed_storage
+            where(arel_table[:storage_version].gteq(HASHED_STORAGE_FEATURES[:repository]))
+          end
+
+          def without_project_repository
+            cond = ProjectRepository.arel_table[:project_id].eq(nil)
+            left_outer_joins(:project_repository).where(cond)
+          end
+
+          def left_outer_joins(relation)
+            return super if Gitlab.rails5?
+
+            # TODO Rails 4?
+          end
+        end
+
+        def project_repository_attributes(shard_finder)
+          return unless hashed_storage?
+
+          {
+            project_id: id,
+            shard_id: shard_finder.find(repository_storage).id,
+            disk_path: hashed_disk_path
+          }
+        end
+
+        private
+
+        def hashed_storage?
+          self.storage_version && self.storage_version >= 1
+        end
+
+        def hashed_disk_path
+          "#{HASHED_PATH_PREFIX}/#{disk_hash[0..1]}/#{disk_hash[2..3]}/#{disk_hash}"
+        end
+
+        def disk_hash
+          @disk_hash ||= Digest::SHA2.hexdigest(id.to_s) if id
+        end
+      end
+
+      def perform(start_id, stop_id)
+        Gitlab::Database.bulk_insert(:project_repositories, project_repositories(start_id, stop_id))
+      end
+
+      private
+
+      def project_repositories(start_id, stop_id)
+        Project.on_hashed_storage.without_project_repository
+          .where(id: start_id..stop_id)
+          .map { |project| project.project_repository_attributes(shard_finder) }
+          .compact
+      end
+
+      def shard_finder
+        @shard_finder ||= ShardFinder.new
+      end
+    end
+  end
+end
diff --git a/spec/lib/gitlab/background_migration/backfill_hashed_project_repositories_spec.rb b/spec/lib/gitlab/background_migration/backfill_hashed_project_repositories_spec.rb
new file mode 100644
index 00000000000..d2f499ffa64
--- /dev/null
+++ b/spec/lib/gitlab/background_migration/backfill_hashed_project_repositories_spec.rb
@@ -0,0 +1,155 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+describe Gitlab::BackgroundMigration::BackfillHashedProjectRepositories, :migration, schema: 20181130102132 do
+  let(:shards) { table(:shards) }
+  let(:namespaces) { table(:namespaces) }
+  let(:projects) { table(:projects) }
+  let(:project_repositories) { table(:project_repositories) }
+  let(:group) { namespaces.create!(name: 'foo', path: 'foo') }
+  let(:default_shard) { shards.create!(name: 'default') }
+
+  describe described_class::ShardFinder do
+    describe '#find' do
+      subject(:finder) { described_class.new }
+
+      it 'creates the shard by name' do
+        expect(finder).to receive(:create!).and_call_original
+
+        expect(finder.find('default')).to be_present
+      end
+
+      it 'does not try to create existing shards' do
+        shards.create(name: 'default')
+
+        expect(finder).not_to receive(:create!)
+
+        finder.find('default')
+      end
+
+      it 'only queries the database once for shards' do
+        finder.find('default')
+
+        expect do
+          finder.find('default')
+        end.not_to exceed_query_limit(0)
+      end
+
+      it 'creates a new shard when it does not exist yet' do
+        expect do
+          finder.find('other')
+        end.to change(shards, :count).by(1)
+      end
+
+      it 'only creates a new shard once' do
+        finder.find('other')
+
+        expect do
+          finder.find('other')
+        end.not_to change(shards, :count)
+      end
+
+      it 'is not vulnerable to race conditions' do
+        finder.find('default')
+
+        other_shard = shards.create(name: 'other')
+
+        expect(finder.find('other').id).to eq(other_shard.id)
+      end
+    end
+  end
+
+  describe described_class::Project do
+    describe '.on_hashed_storage' do
+      it 'finds projects with repository on hashed storage' do
+        hashed_projects = [
+          projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 1),
+          projects.create!(name: 'bar', path: 'bar', namespace_id: group.id, storage_version: 2)
+        ]
+
+        projects.create!(name: 'baz', path: 'baz', namespace_id: group.id, storage_version: 0)
+        projects.create!(name: 'quz', path: 'quz', namespace_id: group.id, storage_version: nil)
+
+        expect(described_class.on_hashed_storage.pluck(:id)).to match_array(hashed_projects.map(&:id))
+      end
+    end
+
+    describe '.without_project_repository' do
+      it 'finds projects which do not have a projects_repositories row' do
+        without_project = projects.create!(name: 'foo', path: 'foo', namespace_id: group.id)
+        with_project = projects.create!(name: 'bar', path: 'bar', namespace_id: group.id)
+        project_repositories.create!(project_id: with_project.id, disk_path: '@phony/foo/bar', shard_id: default_shard.id)
+
+        expect(described_class.without_project_repository.pluck(:id)).to contain_exactly(without_project.id)
+      end
+    end
+
+    describe '#project_repository_attributes' do
+      let(:shard_finder) { Gitlab::BackgroundMigration::BackfillHashedProjectRepositories::ShardFinder.new }
+
+      it 'composes the correct attributes for project_repository' do
+        shiny_shard = shards.create!(name: 'shiny')
+        project = projects.create!(id: 5, name: 'foo', path: 'foo', namespace_id: group.id, repository_storage: shiny_shard.name, storage_version: 1)
+
+        expected_attributes = {
+          project_id: project.id,
+          shard_id: shiny_shard.id,
+          disk_path: '@hashed/ef/2d/ef2d127de37b942baad06145e54b0c619a1f22327b2ebbcfbec78f5564afe39d'
+        }
+
+        expect(described_class.find(project.id).project_repository_attributes(shard_finder)).to eq(expected_attributes)
+      end
+
+      it 'returns nil for a project not on hashed storage' do
+        project = projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 0)
+
+        expect(described_class.find(project.id).project_repository_attributes(shard_finder)).to be_nil
+      end
+    end
+  end
+
+  describe '#perform' do
+    def perform!
+      described_class.new.perform(1, projects.last.id)
+    end
+
+    it 'create project_repository row for hashed storage project' do
+      projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 1)
+
+      expect do
+        perform!
+      end.to change(project_repositories, :count).by(1)
+    end
+
+    it 'does nothing for projects that have already a project_repository' do
+      project = projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 1)
+      project_repositories.create!(project_id: project.id, disk_path: '@phony/foo/bar', shard_id: default_shard.id)
+
+      expect do
+        perform!
+      end.not_to change(project_repositories, :count)
+    end
+
+    it 'does nothing for projects on legacy storage' do
+      projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 0)
+
+      expect do
+        perform!
+      end.not_to change(project_repositories, :count)
+    end
+
+    it 'inserts rows in a single query' do
+      projects.create!(name: 'foo', path: 'foo', namespace_id: group.id, storage_version: 1, repository_storage: default_shard.name)
+
+      control_count = ActiveRecord::QueryRecorder.new do
+        perform!
+      end
+
+      projects.create!(name: 'bar', path: 'bar', namespace_id: group.id, storage_version: 1, repository_storage: default_shard.name)
+      projects.create!(name: 'quz', path: 'quz', namespace_id: group.id, storage_version: 1, repository_storage: default_shard.name)
+
+      expect { perform! }.not_to exceed_query_limit(control_count)
+    end
+  end
+end