Merge branch 'remove-soft-removals' into 'master'

Remove soft removals related code

Closes #37447

See merge request gitlab-org/gitlab-ce!15789

3 files changed, 239 insertions, 8 deletions

diff --git a/db/post_migrate/20171207150343_remove_soft_removed_objects.rb b/db/post_migrate/20171207150343_remove_soft_removed_objects.rb
new file mode 100644
index 00000000000..3e2dedfdd6a
--- /dev/null
+++ b/db/post_migrate/20171207150343_remove_soft_removed_objects.rb
@@ -0,0 +1,208 @@
+# See http://doc.gitlab.com/ce/development/migration_style_guide.html
+# for more information on how to write migrations for GitLab.
+
+class RemoveSoftRemovedObjects < ActiveRecord::Migration
+  include Gitlab::Database::MigrationHelpers
+
+  # Set this constant to true if this migration requires downtime.
+  DOWNTIME = false
+
+  disable_ddl_transaction!
+
+  module SoftRemoved
+    extend ActiveSupport::Concern
+
+    included do
+      scope :soft_removed, -> { where('deleted_at IS NOT NULL') }
+    end
+  end
+
+  class User < ActiveRecord::Base
+    self.table_name = 'users'
+
+    include EachBatch
+  end
+
+  class Issue < ActiveRecord::Base
+    self.table_name = 'issues'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  class MergeRequest < ActiveRecord::Base
+    self.table_name = 'merge_requests'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  class Namespace < ActiveRecord::Base
+    self.table_name = 'namespaces'
+
+    include EachBatch
+    include SoftRemoved
+
+    scope :soft_removed_personal, -> { soft_removed.where(type: nil) }
+    scope :soft_removed_group, -> { soft_removed.where(type: 'Group') }
+  end
+
+  class Route < ActiveRecord::Base
+    self.table_name = 'routes'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  class Project < ActiveRecord::Base
+    self.table_name = 'projects'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  class CiPipelineSchedule < ActiveRecord::Base
+    self.table_name = 'ci_pipeline_schedules'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  class CiTrigger < ActiveRecord::Base
+    self.table_name = 'ci_triggers'
+
+    include EachBatch
+    include SoftRemoved
+  end
+
+  MODELS = [Issue, MergeRequest, CiPipelineSchedule, CiTrigger].freeze
+
+  def up
+    disable_statement_timeout
+
+    remove_personal_routes
+    remove_personal_namespaces
+    remove_group_namespaces
+    remove_simple_soft_removed_rows
+  end
+
+  def down
+    # The data removed by this migration can't be restored in an automated way.
+  end
+
+  def remove_simple_soft_removed_rows
+    create_temporary_indexes
+
+    MODELS.each do |model|
+      say_with_time("Removing soft removed rows from #{model.table_name}") do
+        model.soft_removed.each_batch do |batch, index|
+          batch.delete_all
+        end
+      end
+    end
+  ensure
+    remove_temporary_indexes
+  end
+
+  def create_temporary_indexes
+    MODELS.each do |model|
+      index_name = temporary_index_name_for(model)
+
+      # Without this index the removal process can take a very long time. For
+      # example, getting the next ID of a batch for the `issues` table in
+      # staging would take between 15 and 20 seconds.
+      next if temporary_index_exists?(model)
+
+      say_with_time("Creating temporary index #{index_name}") do
+        add_concurrent_index(
+          model.table_name,
+          [:deleted_at, :id],
+          name: index_name,
+          where: 'deleted_at IS NOT NULL'
+        )
+      end
+    end
+  end
+
+  def remove_temporary_indexes
+    MODELS.each do |model|
+      index_name = temporary_index_name_for(model)
+
+      next unless temporary_index_exists?(model)
+
+      say_with_time("Removing temporary index #{index_name}") do
+        remove_concurrent_index_by_name(model.table_name, index_name)
+      end
+    end
+  end
+
+  def temporary_index_name_for(model)
+    "index_on_#{model.table_name}_tmp"
+  end
+
+  def temporary_index_exists?(model)
+    index_name = temporary_index_name_for(model)
+
+    index_exists?(model.table_name, [:deleted_at, :id], name: index_name)
+  end
+
+  def remove_personal_namespaces
+    # Some personal namespaces are left behind in case of GitLab.com. In these
+    # cases the associated data such as the projects and users has already been
+    # removed.
+    Namespace.soft_removed_personal.each_batch do |batch|
+      batch.delete_all
+    end
+  end
+
+  def remove_group_namespaces
+    admin_id = id_for_admin_user
+
+    unless admin_id
+      say 'Not scheduling soft removed groups for removal as no admin user ' \
+        'could be found. You will need to remove any such groups manually.'
+
+      return
+    end
+
+    # Left over groups can't be easily removed because we may also need to
+    # remove memberships, repositories, and other associated data. As a result
+    # we'll just schedule a Sidekiq job to remove these.
+    #
+    # As of January 5th, 2018 there are 36 groups that will be removed using
+    # this code.
+    Namespace.select(:id).soft_removed_group.each_batch(of: 10) do |batch, index|
+      batch.each do |ns|
+        schedule_group_removal(index * 5.minutes, ns.id, admin_id)
+      end
+    end
+  end
+
+  def schedule_group_removal(delay, group_id, user_id)
+    if migrate_inline?
+      GroupDestroyWorker.new.perform(group_id, user_id)
+    else
+      GroupDestroyWorker.perform_in(delay, group_id, user_id)
+    end
+  end
+
+  def remove_personal_routes
+    namespaces = Namespace.select(1)
+      .soft_removed
+      .where('namespaces.type IS NULL')
+      .where('routes.source_type = ?', 'Namespace')
+      .where('routes.source_id = namespaces.id')
+
+    Route.where('EXISTS (?)', namespaces).each_batch do |batch|
+      batch.delete_all
+    end
+  end
+
+  def id_for_admin_user
+    User.where(admin: true).limit(1).pluck(:id).first
+  end
+
+  def migrate_inline?
+    Rails.env.test? || Rails.env.development?
+  end
+end
diff --git a/db/post_migrate/20171207150344_remove_deleted_at_columns.rb b/db/post_migrate/20171207150344_remove_deleted_at_columns.rb
new file mode 100644
index 00000000000..154d7a1b926
--- /dev/null
+++ b/db/post_migrate/20171207150344_remove_deleted_at_columns.rb
@@ -0,0 +1,31 @@
+# See http://doc.gitlab.com/ce/development/migration_style_guide.html
+# for more information on how to write migrations for GitLab.
+
+class RemoveDeletedAtColumns < ActiveRecord::Migration
+  include Gitlab::Database::MigrationHelpers
+
+  DOWNTIME = false
+
+  disable_ddl_transaction!
+
+  TABLES = %i[issues merge_requests namespaces ci_pipeline_schedules ci_triggers].freeze
+  COLUMN = :deleted_at
+
+  def up
+    TABLES.each do |table|
+      remove_column(table, COLUMN) if column_exists?(table, COLUMN)
+    end
+  end
+
+  def down
+    TABLES.each do |table|
+      unless column_exists?(table, COLUMN)
+        add_column(table, COLUMN, :datetime_with_timezone)
+      end
+
+      unless index_exists?(table, COLUMN)
+        add_concurrent_index(table, COLUMN)
+      end
+    end
+  end
+end
diff --git a/db/schema.rb b/db/schema.rb
index 07620349057..8a6db61250b 100644
--- a/db/schema.rb
+++ b/db/schema.rb
@@ -357,7 +357,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
     t.integer "project_id"
     t.integer "owner_id"
     t.boolean "active", default: true
-    t.datetime "deleted_at"
     t.datetime "created_at"
     t.datetime "updated_at"
   end
@@ -467,7 +466,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
 
   create_table "ci_triggers", force: :cascade do |t|
     t.string "token"
-    t.datetime "deleted_at"
     t.datetime "created_at"
     t.datetime "updated_at"
     t.integer "project_id"
@@ -861,7 +859,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
     t.integer "iid"
     t.integer "updated_by_id"
     t.boolean "confidential", default: false, null: false
-    t.datetime "deleted_at"
     t.date "due_date"
     t.integer "moved_to_id"
     t.integer "lock_version"
@@ -878,7 +875,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
 
   add_index "issues", ["author_id"], name: "index_issues_on_author_id", using: :btree
   add_index "issues", ["confidential"], name: "index_issues_on_confidential", using: :btree
-  add_index "issues", ["deleted_at"], name: "index_issues_on_deleted_at", using: :btree
   add_index "issues", ["description"], name: "index_issues_on_description_trigram", using: :gin, opclasses: {"description"=>"gin_trgm_ops"}
   add_index "issues", ["milestone_id"], name: "index_issues_on_milestone_id", using: :btree
   add_index "issues", ["moved_to_id"], name: "index_issues_on_moved_to_id", where: "(moved_to_id IS NOT NULL)", using: :btree
@@ -1088,7 +1084,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
     t.boolean "merge_when_pipeline_succeeds", default: false, null: false
     t.integer "merge_user_id"
     t.string "merge_commit_sha"
-    t.datetime "deleted_at"
     t.string "in_progress_merge_commit_sha"
     t.integer "lock_version"
     t.text "title_html"
@@ -1107,7 +1102,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
   add_index "merge_requests", ["assignee_id"], name: "index_merge_requests_on_assignee_id", using: :btree
   add_index "merge_requests", ["author_id"], name: "index_merge_requests_on_author_id", using: :btree
   add_index "merge_requests", ["created_at"], name: "index_merge_requests_on_created_at", using: :btree
-  add_index "merge_requests", ["deleted_at"], name: "index_merge_requests_on_deleted_at", using: :btree
   add_index "merge_requests", ["description"], name: "index_merge_requests_on_description_trigram", using: :gin, opclasses: {"description"=>"gin_trgm_ops"}
   add_index "merge_requests", ["head_pipeline_id"], name: "index_merge_requests_on_head_pipeline_id", using: :btree
   add_index "merge_requests", ["latest_merge_request_diff_id"], name: "index_merge_requests_on_latest_merge_request_diff_id", using: :btree
@@ -1167,7 +1161,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
     t.boolean "share_with_group_lock", default: false
     t.integer "visibility_level", default: 20, null: false
     t.boolean "request_access_enabled", default: false, null: false
-    t.datetime "deleted_at"
     t.text "description_html"
     t.boolean "lfs_enabled"
     t.integer "parent_id"
@@ -1177,7 +1170,6 @@ ActiveRecord::Schema.define(version: 20180105212544) do
   end
 
   add_index "namespaces", ["created_at"], name: "index_namespaces_on_created_at", using: :btree
-  add_index "namespaces", ["deleted_at"], name: "index_namespaces_on_deleted_at", using: :btree
   add_index "namespaces", ["name", "parent_id"], name: "index_namespaces_on_name_and_parent_id", unique: true, using: :btree
   add_index "namespaces", ["name"], name: "index_namespaces_on_name_trigram", using: :gin, opclasses: {"name"=>"gin_trgm_ops"}
   add_index "namespaces", ["owner_id"], name: "index_namespaces_on_owner_id", using: :btree