Seed data for users, projects and its relations

It handles the very basics for batch inserting
records using the seed database, also showing
time estimatives.

It leverages generate_series() (from postgres) and
Gitlab::Database.bulk_insert methods for
inserting a great amount of users, projects, routes
and namespaces.

Gemfile

@@ -336,7 +336,6 @@ group :development do
 
   gem 'letter_opener_web', '~> 1.3.4'
   gem 'rblineprof', '~> 0.3.6', platform: :mri, require: false
-  gem 'active_record-pg_generate_series', '~> 0.1.2'
 
   # Better errors handler
   gem 'better_errors', '~> 2.5.0'

Gemfile.lock

@@ -29,8 +29,6 @@ GEM
       erubi (~> 1.4)
       rails-dom-testing (~> 2.0)
       rails-html-sanitizer (~> 1.0, >= 1.0.3)
-    active_record-pg_generate_series (0.1.3)
-      activerecord
     activejob (5.2.3)
       activesupport (= 5.2.3)
       globalid (>= 0.3.6)
@@ -1090,7 +1088,6 @@ DEPENDENCIES
   RedCloth (~> 4.3.2)
   ace-rails-ap (~> 4.1.0)
   acme-client (~> 2.0.2)
-  active_record-pg_generate_series (~> 0.1.2)
   activerecord-explain-analyze (~> 0.1)
   acts-as-taggable-on (~> 6.0)
   addressable (~> 2.5.2)

db/fixtures/development/02_users.rb

+# frozen_string_literal: true
+
 class Gitlab::Seeder::Users
   include ActionView::Helpers::NumberHelper
 
   RANDOM_USERS_COUNT = 20
-  MASS_USERS_COUNT = 1_500_000
+  MASS_USERS_COUNT = ENV['CI'] ? 10 : 1_000_000
+  MASS_INSERT_USERNAME_START = 'mass_insert_user_'
 
   attr_reader :opts
 
@@ -12,13 +15,43 @@ class Gitlab::Seeder::Users
 
   def seed!
     Sidekiq::Testing.inline! do
-      create_random_users!
       create_mass_users!
+      create_random_users!
     end
   end
 
   private
 
+  def create_mass_users!
+    encrypted_password = Devise::Encryptor.digest(User, '12345678')
+
+    Gitlab::Seeder.with_mass_insert(MASS_USERS_COUNT, User) do
+      ActiveRecord::Base.connection.execute <<~SQL
+        INSERT INTO users (username, name, email, confirmed_at, projects_limit, encrypted_password)
+        SELECT
+          '#{MASS_INSERT_USERNAME_START}' || seq,
+          'Seed user ' || seq,
+          'seed_user' || seq || '@example.com',
+          to_timestamp(seq),
+          #{MASS_USERS_COUNT},
+          '#{encrypted_password}'
+        FROM generate_series(1, #{MASS_USERS_COUNT}) AS seq
+      SQL
+    end
+
+    relation = User.where(admin: false)
+    Gitlab::Seeder.with_mass_insert(relation.count, Namespace) do
+      ActiveRecord::Base.connection.execute <<~SQL
+        INSERT INTO namespaces (name, path, owner_id)
+        SELECT
+          username,
+          username,
+          id
+        FROM users WHERE NOT admin
+      SQL
+    end
+  end
+
   def create_random_users!
     RANDOM_USERS_COUNT.times do |i|
       begin
@@ -36,26 +69,6 @@ class Gitlab::Seeder::Users
       end
     end
   end
-
-  def create_mass_users!
-    # Disable database insertion logs so speed isn't limited by ability to print to console
-    old_logger = ActiveRecord::Base.logger
-    ActiveRecord::Base.logger = nil
-
-    encrypted_password = Devise::Encryptor.digest(User, '12345678')
-
-    User.insert_using_generate_series(1, MASS_USERS_COUNT, debug: true) do |sql|
-      sql.username = raw("'user' || seq")
-      sql.name = raw("'User ' || seq")
-      sql.email = raw("'user' || seq || '@example.com'")
-      sql.confirmed_at = raw("('1388530801'::timestamp + seq)::date") # 2014-01-01
-      sql.encrypted_password = encrypted_password
-    end
-    puts "\n#{number_with_delimiter(MASS_USERS_COUNT)} users created!"
-
-    # Reset logging
-    ActiveRecord::Base.logger = old_logger
-  end
 end
 
 Gitlab::Seeder.quiet do

db/fixtures/development/04_labels.rb

@@ -43,7 +43,7 @@ Gitlab::Seeder.quiet do
   end
 
   puts "\nGenerating project labels"
-  Project.all.find_each do |project|
+  Project.not_mass_generated.find_each do |project|
     Gitlab::Seeder::ProjectLabels.new(project).seed!
   end
 end

db/fixtures/development/04_projects.rb

-require './spec/support/sidekiq'
-
-class Gitlab::Seeder::Projects
-  include ActionView::Helpers::NumberHelper
-
-  PROJECT_URLS = [
-    'https://gitlab.com/gitlab-org/gitlab-test.git',
-    'https://gitlab.com/gitlab-org/gitlab-ce.git',
-    'https://gitlab.com/gitlab-org/gitlab-ci.git',
-    'https://gitlab.com/gitlab-org/gitlab-shell.git',
-    'https://github.com/documentcloud/underscore.git',
-    'https://github.com/twitter/flight.git',
-    'https://github.com/twitter/typeahead.js.git',
-    'https://github.com/h5bp/html5-boilerplate.git',
-    'https://github.com/google/material-design-lite.git',
-    'https://github.com/jlevy/the-art-of-command-line.git',
-    'https://github.com/FreeCodeCamp/freecodecamp.git',
-    'https://github.com/google/deepdream.git',
-    'https://github.com/jtleek/datasharing.git',
-    'https://github.com/WebAssembly/design.git',
-    'https://github.com/airbnb/javascript.git',
-    'https://github.com/tessalt/echo-chamber-js.git',
-    'https://github.com/atom/atom.git',
-    'https://github.com/mattermost/platform.git',
-    'https://github.com/purifycss/purifycss.git',
-    'https://github.com/facebook/nuclide.git',
-    'https://github.com/wbkd/awesome-d3.git',
-    'https://github.com/kilimchoi/engineering-blogs.git',
-    'https://github.com/gilbarbara/logos.git',
-    'https://github.com/gaearon/redux.git',
-    'https://github.com/awslabs/s2n.git',
-    'https://github.com/arkency/reactjs_koans.git',
-    'https://github.com/twbs/bootstrap.git',
-    'https://github.com/chjj/ttystudio.git',
-    'https://github.com/DrBoolean/mostly-adequate-guide.git',
-    'https://github.com/octocat/Spoon-Knife.git',
-    'https://github.com/opencontainers/runc.git',
-    'https://github.com/googlesamples/android-topeka.git'
-  ]
-  MASS_PROJECTS_COUNT = {
-    private: 2_000_000,
-    internal: 30_000,
-    public: 265_000
-  }
-
-  attr_reader :opts
-
-  def initialize(opts = {})
-    @opts = opts
-  end
-
-  def seed!
-    Sidekiq::Testing.inline! do
-      create_real_projects!(opts[:count])
-      create_mass_projects!
-    end
-  end
-
-  private
-
-  def create_real_projects!(count)
-    PROJECT_URLS.first(count).each_with_index do |url, i|
-      group_path, project_path = url.split('/')[-2..-1]
-
-      group = Group.find_by(path: group_path)
-
-      unless group
-        group = Group.new(
-          name: group_path.titleize,
-          path: group_path
-        )
-        group.description = FFaker::Lorem.sentence
-        group.save
-
-        group.add_owner(User.first)
-      end
-
-      project_path.gsub!(".git", "")
-
-      params = {
-        import_url: url,
-        namespace_id: group.id,
-        name: project_path.titleize,
-        description: FFaker::Lorem.sentence,
-        visibility_level: Gitlab::VisibilityLevel.values.sample
-      }
-
-      project = ::Projects::CreateService.new(User.first, params).execute
-      # Seed-Fu runs this entire fixture in a transaction, so the `after_commit`
-      # hook won't run until after the fixture is loaded. That is too late
-      # since the Sidekiq::Testing block has already exited. Force clearing
-      # the `after_commit` queue to ensure the job is run now.
-      project.send(:_run_after_commit_queue)
-
-      if project.valid? && project.valid_repo?
-        print '.'
-      else
-        puts project.errors.full_messages
-        print 'F'
-      end
-    end
-  end
-
-  def create_mass_projects!
-    # Disable database insertion logs so speed isn't limited by ability to print to console
-    old_logger = ActiveRecord::Base.logger
-    ActiveRecord::Base.logger = nil
-
-    create_mass_projects_by_visility!(:private)
-    create_mass_projects_by_visility!(:internal)
-    create_mass_projects_by_visility!(:public)
-
-    # Reset logging
-    ActiveRecord::Base.logger = old_logger
-  end
-
-  def create_mass_projects_by_visility!(visibility)
-    users = User.limit(100)
-    groups = Group.limit(100)
-    namespaces = users + groups
-    Project.insert_using_generate_series(1, MASS_PROJECTS_COUNT[visibility], debug: true) do |sql|
-      project_name = raw("'seed_#{visibility}_project_' || seq")
-      namespace = namespaces.take
-      sql.name = project_name
-      sql.path = project_name
-      sql.creator_id = namespace.is_a?(Group) ? namespace.owner_id : users.take.id
-      sql.namespace_id = namespace.is_a?(Group) ? namespace.id : namespace.namespace_id
-      sql.visibility_level = Gitlab::VisibilityLevel.level_value(visibility.to_s)
-    end
-    puts "#{number_with_delimiter(MASS_PROJECTS_COUNT[visibility])} projects created!"
-  end
-end
-
-Gitlab::Seeder.quiet do
-  count = ENV['SIZE'].present? ? ENV['SIZE'].to_i : 8
-  projects = Gitlab::Seeder::Projects.new(count: count)
-  projects.seed!
-end

db/fixtures/development/06_teams.rb

@@ -3,7 +3,7 @@ require './spec/support/sidekiq'
 Sidekiq::Testing.inline! do
   Gitlab::Seeder.quiet do
     Group.all.each do |group|
-      User.all.sample(4).each do |user|
+      User.not_mass_generated.sample(4).each do |user|
         if group.add_user(user, Gitlab::Access.values.sample).persisted?
           print '.'
         else
@@ -12,8 +12,8 @@ Sidekiq::Testing.inline! do
       end
     end
 
-    Project.all.each do |project|
-      User.all.sample(4).each do |user|
+    Project.not_mass_generated.each do |project|
+      User.not_mass_generated.sample(4).each do |user|
         if project.add_role(user, Gitlab::Access.sym_options.keys.sample)
           print '.'
         else

db/fixtures/development/07_milestones.rb

 require './spec/support/sidekiq'
 
 Gitlab::Seeder.quiet do
-  Project.all.each do |project|
+  Project.not_mass_generated.each do |project|
     5.times do |i|
       milestone_params = {
         title: "v#{i}.0",

db/fixtures/development/10_merge_requests.rb

@@ -4,7 +4,13 @@ Gitlab::Seeder.quiet do
   # Limit the number of merge requests per project to avoid long seeds
   MAX_NUM_MERGE_REQUESTS = 10
 
-  Project.non_archived.with_merge_requests_enabled.reject(&:empty_repo?).each do |project|
+  projects = Project
+    .non_archived
+    .with_merge_requests_enabled
+    .not_mass_generated
+    .reject(&:empty_repo?)
+
+  projects.each do |project|
     branches = project.repository.branch_names.sample(MAX_NUM_MERGE_REQUESTS * 2)
 
     branches.each do |branch_name|

db/fixtures/development/11_keys.rb

@@ -9,7 +9,7 @@ Sidekiq::Testing.disable! do
     # that it falls under `Sidekiq::Testing.disable!`.
     Key.skip_callback(:commit, :after, :add_to_shell)
 
-    User.first(10).each do |user|
+    User.not_mass_generated.first(10).each do |user|
       key = "ssh-rsa AAAAB3NzaC1yc2EAAAABJQAAAIEAiPWx6WM4lhHNedGfBpPJNPpZ7yKu+dnn1SJejgt#{user.id + 100}6k6YjzGGphH2TUxwKzxcKDKKezwkpfnxPkSMkuEspGRt/aZZ9wa++Oi7Qkr8prgHc4soW6NUlfDzpvZK2H5E7eQaSeP3SAwGmQKUFHCddNaP0L+hM7zhFNzjFvpaMgJw0="
 
       key = user.keys.create(

db/fixtures/development/12_snippets.rb

@@ -25,7 +25,7 @@ end
 eos
 
   50.times do |i|
-    user = User.all.sample
+    user = User.not_mass_generated.sample
 
     PersonalSnippet.seed(:id, [{
       id: i,

db/fixtures/development/14_pipelines.rb

@@ -214,7 +214,7 @@ class Gitlab::Seeder::Pipelines
 end
 
 Gitlab::Seeder.quiet do
-  Project.all.sample(5).each do |project|
+  Project.not_mass_generated.sample(5).each do |project|
     project_builds = Gitlab::Seeder::Pipelines.new(project)
     project_builds.seed!
   end

db/fixtures/development/16_protected_branches.rb

@@ -3,7 +3,7 @@ require './spec/support/sidekiq'
 Gitlab::Seeder.quiet do
   admin_user = User.find(1)
 
-  Project.all.each do |project|
+  Project.not_mass_generated.each do |project|
     params = {
       name: 'master'
     }

db/fixtures/development/17_cycle_analytics.rb

@@ -217,7 +217,7 @@ Gitlab::Seeder.quiet do
   flag = 'SEED_CYCLE_ANALYTICS'
 
   if ENV[flag]
-    Project.find_each do |project|
+    Project.not_mass_generated.find_each do |project|
       # This seed naively assumes that every project has a repository, and every
       # repository has a `master` branch, which may be the case for a pristine
       # GDK seed, but is almost never true for a GDK that's actually had

db/fixtures/development/19_environments.rb

@@ -67,7 +67,7 @@ class Gitlab::Seeder::Environments
 end
 
 Gitlab::Seeder.quiet do
-  Project.all.sample(5).each do |project|
+  Project.not_mass_generated.sample(5).each do |project|
     project_environments = Gitlab::Seeder::Environments.new(project)
     project_environments.seed!
   end

db/fixtures/development/23_spam_logs.rb

@@ -22,7 +22,7 @@ module Db
         end
 
         def self.random_user
-          User.find(User.pluck(:id).sample)
+          User.find(User.not_mass_generated.pluck(:id).sample)
         end
       end
     end

db/fixtures/development/24_forks.rb

@@ -2,8 +2,8 @@ require './spec/support/sidekiq'
 
 Sidekiq::Testing.inline! do
   Gitlab::Seeder.quiet do
-    User.all.sample(10).each do |user|
-      source_project = Project.public_only.sample
+    User.not_mass_generated.sample(10).each do |user|
+      source_project = Project.not_mass_generated.public_only.sample
 
       ##
       # 03_project.rb might not have created a public project because

doc/development/rake_tasks.md

@@ -12,6 +12,14 @@ The `setup` task is an alias for `gitlab:setup`.
 This tasks calls `db:reset` to create the database, and calls `db:seed_fu` to seed the database.
 Note: `db:setup` calls `db:seed` but this does nothing.
 
+### Env variables
+
+**MASS_INSERT**: Create millions of users (2m), projects (5m) and its
+relations. It's highly recommended to run the seed with it to catch slow queries
+while developing. Expect the process to take up to 20 extra minutes.
+
+**LARGE_PROJECTS**: Create large projects (through import) from a predefined set of urls.
+
 ### Seeding issues for all or a given project
 
 You can seed issues for all or a given project with the `gitlab:seed:issues`

ee/db/fixtures/development/20_burndown.rb

@@ -88,7 +88,7 @@ Gitlab::Seeder.quiet do
     seeder = Gitlab::Seeder::Burndown.new(project)
     seeder.seed!
   else
-    Project.all.each do |project|
+    Project.not_mass_generated.each do |project|
       seeder = Gitlab::Seeder::Burndown.new(project)
       seeder.seed!
     end

ee/db/fixtures/development/20_vulnerabilities.rb

@@ -128,7 +128,7 @@ class Gitlab::Seeder::Vulnerabilities
 end
 
 Gitlab::Seeder.quiet do
-  Project.joins(:ci_pipelines).distinct.all.sample(5).each do |project|
+  Project.joins(:ci_pipelines).not_mass_generated.distinct.all.sample(5).each do |project|
     seeder = Gitlab::Seeder::Vulnerabilities.new(project)
     seeder.seed!
   end

ee/db/fixtures/development/25_downstream_pipelines.rb

 # frozen_string_literal: true
 # EE fixture
 Gitlab::Seeder.quiet do
-  Project.all.sample(5).each do |project|
+  Project.not_mass_generated.sample(5).each do |project|
     project.ci_pipelines.all.sample(2).each do |pipeline|
       next if pipeline.source_pipeline
 

ee/db/fixtures/development/26_packages.rb

@@ -32,7 +32,7 @@ class Gitlab::Seeder::Packages
 end
 
 Gitlab::Seeder.quiet do
-  Project.all.sample(5).each do |project|
+  Project.not_mass_generated.sample(5).each do |project|
     Gitlab::Seeder::Packages.new(project.owner, project).seed
   end
 end

lib/gitlab/seeder.rb

@@ -14,7 +14,71 @@ end
 
 module Gitlab
   class Seeder
+    extend ActionView::Helpers::NumberHelper
+
+    ESTIMATED_INSERT_PER_MINUTE = 2_000_000
+    MASS_INSERT_ENV = 'MASS_INSERT'
+
+    module ProjectSeed
+      extend ActiveSupport::Concern
+
+      included do
+        scope :not_mass_generated, -> do
+          where.not("path LIKE '#{Gitlab::Seeder::Projects::MASS_INSERT_NAME_START}%'")
+        end
+      end
+    end
+
+    module UserSeed
+      extend ActiveSupport::Concern
+
+      included do
+        scope :not_mass_generated, -> do
+          where.not("username LIKE '#{Gitlab::Seeder::Users::MASS_INSERT_USERNAME_START}%'")
+        end
+      end
+    end
+
+    def self.with_mass_insert(size, model)
+      humanized_model_name = model.is_a?(String) ? model : model.model_name.human.pluralize(size)
+
+      if !ENV[MASS_INSERT_ENV] && !ENV['CI']
+        puts "\nSkipping mass insertion for #{humanized_model_name}."
+        puts "Consider running the seed with #{MASS_INSERT_ENV}=1"
+        return
+      end
+
+      humanized_size = number_with_delimiter(size)
+      estimative = estimated_time_message(size)
+
+      puts "\nCreating #{humanized_size} #{humanized_model_name}."
+      puts estimative
+
+      yield
+
+      puts "\n#{number_with_delimiter(size)} #{humanized_model_name} created!"
+    end
+
+    def self.estimated_time_message(size)
+      estimated_minutes = (size.to_f / ESTIMATED_INSERT_PER_MINUTE).round
+      humanized_minutes = 'minute'.pluralize(estimated_minutes)
+
+      if estimated_minutes.zero?
+        "Rough estimated time: less than a minute ⏰"
+      else
+        "Rough estimated time: #{estimated_minutes} #{humanized_minutes} ⏰"
+      end
+    end
+
     def self.quiet
+      # Disable database insertion logs so speed isn't limited by ability to print to console
+      old_logger = ActiveRecord::Base.logger
+      ActiveRecord::Base.logger = nil
+
+      # Additional seed logic for models.
+      Project.include(ProjectSeed)
+      User.include(UserSeed)
+
       mute_notifications
       mute_mailer
 
@@ -23,6 +87,7 @@ module Gitlab
       yield
 
       SeedFu.quiet = false
+      ActiveRecord::Base.logger = old_logger
       puts "\nOK".color(:green)
     end
 

lib/tasks/dev.rake

@@ -5,6 +5,10 @@ namespace :dev do
   task setup: :environment do
     ENV['force'] = 'yes'
     Rake::Task["gitlab:setup"].invoke
+
+    # Make sure DB statistics are up to date.
+    ActiveRecord::Base.connection.execute('ANALYZE')
+
     Rake::Task["gitlab:shell:setup"].invoke
   end
 

lib/tasks/gitlab/seed.rake

@@ -22,7 +22,7 @@ namespace :gitlab do
 
           [project]
         else
-          Project.find_each
+          Project.not_mass_generated.find_each
         end
 
       projects.each do |project|