Merge branch 'feature/geo-improvements' into 'master'

Geo: Improvements and fixes after QA

We are going to move Geo (#76) to General Availability. This MR will handle all small fixes and changes to get a solid release

- [x] Replicates repository removal (using system hooks)
- [x] Replicates repository rename (using system hooks)
- [x] Replicates repository transfer (using system hooks)
- [x] Whitelisted Sidekiq routes so you can use web interface in a secondary node.
- [x] Change "Gitlab -> GitLab" for the readonly error flash
- [x] Check license add-on to allow any Geo features. 
- [x] Do not execute on a secondary node: `StuckCiBuildsWorker`
- [x] Do not execute on a secondary node: `HistoricalDataWorker`
- [x] Documentation changes: (master, slave -> primary, secondary)
- [x] Document we require `db_key_base`(secrets.yml) to be the same in all nodes

See merge request !354

CHANGELOG-EE

@@ -5,6 +5,7 @@ v 8.7.0
   - Refactor group sync to pull access level logic to its own class. !306
   - [Elastic] Stabilize database indexer if database is inconsistent
   - Add ability to sync to remote mirrors. !249
+  - GitLab Geo: Many replication improvements and fixes !354
 
 v 8.6.7
   - No EE-specific changes

app/controllers/admin/geo_nodes_controller.rb

 class Admin::GeoNodesController < Admin::ApplicationController
+  before_action :check_license
+
   def index
     @nodes = GeoNode.all
     @node = GeoNode.new
@@ -41,4 +43,11 @@ class Admin::GeoNodesController < Admin::ApplicationController
   def geo_node_params
     params.require(:geo_node).permit(:url, :primary, geo_node_key_attributes: [:key])
   end
+
+  def check_license
+    unless Gitlab::Geo.license_allows?
+      flash[:alert] = 'You need a diferent license to enable Geo replication'
+      redirect_to admin_license_path
+    end
+  end
 end

app/services/geo/move_repository_service.rb

+module Geo
+  class MoveRepositoryService
+    include Gitlab::ShellAdapter
+
+    attr_reader :id, :name, :old_path_with_namespace, :new_path_with_namespace
+
+    def initialize(id, name, old_path_with_namespace, new_path_with_namespace)
+      @id = id
+      @name = name
+      @old_path_with_namespace = old_path_with_namespace
+      @new_path_with_namespace = new_path_with_namespace
+    end
+
+    def execute
+      project = Project.find(id)
+      project.expire_caches_before_rename(old_path_with_namespace)
+
+      # Make sure target directory exists (used when transfering repositories)
+      project.namespace.ensure_dir_exist
+
+      if gitlab_shell.mv_repository(old_path_with_namespace, new_path_with_namespace)
+        # If repository moved successfully we need to send update instructions to users.
+        # However we cannot allow rollback since we moved repository
+        # So we basically we mute exceptions in next actions
+        begin
+          gitlab_shell.mv_repository("#{old_path_with_namespace}.wiki", "#{new_path_with_namespace}.wiki")
+        rescue
+          # Returning false does not rollback after_* transaction but gives
+          # us information about failing some of tasks
+          false
+        end
+      else
+        # if we cannot move namespace directory we should rollback
+        # db changes in order to prevent out of sync between db and fs
+        raise Exception.new('repository cannot be renamed')
+      end
+    end
+  end
+end

app/services/geo/schedule_repo_destroy_service.rb

+module Geo
+  class ScheduleRepoDestroyService
+    attr_reader :id, :name, :path_with_namespace
+
+    def initialize(params)
+      @id = params['project_id']
+      @name = params['name']
+      @path_with_namespace = params['path_with_namespace']
+    end
+
+    def execute
+      GeoRepositoryDestroyWorker.perform_async(id, name, path_with_namespace)
+    end
+  end
+end

app/services/geo/schedule_repo_move_service.rb

+module Geo
+  class ScheduleRepoRenameService
+    attr_reader :id, :name, :old_path_with_namespace, :path_with_namespace
+
+    def initialize(params)
+      @id = params['project_id']
+      @name = params['name']
+      @old_path_with_namespace = params['old_path_with_namespace']
+      @path_with_namespace = params['path_with_namespace']
+    end
+
+    def execute
+      GeoRepositoryMoveWorker.perform_async(id, name, old_path_with_namespace, path_with_namespace)
+    end
+  end
+end

app/services/projects/destroy_service.rb

@@ -17,9 +17,6 @@ module Projects
 
       project.team.truncate
 
-      repo_path = project.path_with_namespace
-      wiki_path = repo_path + '.wiki'
-
       # Flush the cache for both repositories. This has to be done _before_
       # removing the physical repositories as some expiration code depends on
       # Git data (e.g. a list of branch names).
@@ -27,14 +24,7 @@ module Projects
 
       Project.transaction do
         project.destroy!
-
-        unless remove_repository(repo_path)
-          raise_error('Failed to remove project repository. Please try again or contact administrator')
-        end
-
-        unless remove_repository(wiki_path)
-          raise_error('Failed to remove wiki repository. Please try again or contact administrator')
-        end
+        trash_repositories!
       end
 
       log_info("Project \"#{project.name}\" was removed")
@@ -42,8 +32,39 @@ module Projects
       true
     end
 
+    # Removes physical repository in a Geo replicated secondary node
+    # There is no need to do any database operation as it will be
+    # replicated by itself.
+    def geo_replicate
+      # Flush the cache for both repositories. This has to be done _before_
+      # removing the physical repositories as some expiration code depends on
+      # Git data (e.g. a list of branch names).
+      flush_caches(project, wiki_path)
+
+      trash_repositories!
+      log_info("Project \"#{project.name}\" was removed")
+    end
+
     private
 
+    def repo_path
+      project.path_with_namespace
+    end
+
+    def wiki_path
+      repo_path + '.wiki'
+    end
+
+    def trash_repositories!
+      unless remove_repository(repo_path)
+        raise_error('Failed to remove project repository. Please try again or contact administrator')
+      end
+
+      unless remove_repository(wiki_path)
+        raise_error('Failed to remove wiki repository. Please try again or contact administrator')
+      end
+    end
+
     def remove_repository(path)
       # Skip repository removal. We use this flag when remove user or group
       return true if params[:skip_repo] == true

app/workers/geo_repository_destroy_worker.rb

+class GeoRepositoryDestroyWorker
+  include Sidekiq::Worker
+
+  sidekiq_options queue: :default
+
+  def perform(id, name, path_with_namespace)
+    # We don't have access to the original model anymore, so we are
+    # rebuilding only what our service class requires
+    project = FakeProject.new(id, name, path_with_namespace)
+
+    ::Projects::DestroyService.new(project, nil).geo_replicate
+  end
+
+  FakeProject = Struct.new(:id, :name, :path_with_namespace) do
+    def repository
+      @repository ||= Repository.new(path_with_namespace, self)
+    end
+  end
+end

app/workers/geo_repository_move_worker.rb

+class GeoRepositoryMoveWorker
+  include Sidekiq::Worker
+
+  sidekiq_options queue: :default
+
+  def perform(id, name, old_path_with_namespace, new_path_with_namespace)
+    Geo::MoveRepositoryService.new(id, name, old_path_with_namespace, new_path_with_namespace).execute
+  end
+end

app/workers/historical_data_worker.rb

@@ -2,6 +2,7 @@ class HistoricalDataWorker
   include Sidekiq::Worker
 
   def perform
+    return if Gitlab::Geo.secondary?
     HistoricalData.track!
   end
 end

app/workers/stuck_ci_builds_worker.rb

@@ -4,6 +4,7 @@ class StuckCiBuildsWorker
   BUILD_STUCK_TIMEOUT = 1.day
 
   def perform
+    return if Gitlab::Geo.secondary?
     Rails.logger.info 'Cleaning stuck builds'
 
     builds = Ci::Build.running_or_pending.where('updated_at < ?', BUILD_STUCK_TIMEOUT.ago)

doc/gitlab-geo/README.md

@@ -54,7 +54,7 @@ Geo instances. Follow the steps below in the order that they appear:
 
 1. Install GitLab Enterprise Edition on the server that will serve as the
    secondary Geo node
-1. [Setup a database replication](./database.md) in `master <-> slave` topology
+1. [Setup a database replication](./database.md) in `primary <-> secondary (read-only)` topology
 1. [Configure GitLab](configuration.md) and set the primary and secondary nodes
 After you set up the database replication and configure the GitLab Geo nodes,
 there are a few things to consider:

doc/gitlab-geo/configuration.md

@@ -14,7 +14,8 @@ complete the process.
 - [Create SSH key pairs for Geo nodes](#create-ssh-key-pairs-for-geo-nodes)
 - [Primary Node GitLab setup](#primary-node-gitlab-setup)
 - [Secondary Node GitLab setup](#secondary-node-gitlab-setup)
-    - [Authorized keys regeneration](#authorized-keys-regeneration)
+  - [Database Encryptation Key](#database-encryptation-key)
+  - [Authorized keys regeneration](#authorized-keys-regeneration)
 
 <!-- END doctoc generated TOC please keep comment here to allow auto update -->
 
@@ -81,8 +82,8 @@ Host example.com                    # The FQDN of the primary Geo node
 ## Primary Node GitLab setup
 
 >**Note:**
-You will need to setup your database into a **Master <-> Slave** replication
-topology, and your Primary node should always point to a database's Master
+You will need to setup your database into a **Primary <-> Secondary (read-only)** replication
+topology, and your Primary node should always point to a database's Primary
 instance. If you haven't done that already, read [database replication](./database.md).
 
 Go to the server that you chose to be your primary, and visit
@@ -119,6 +120,18 @@ Edition installation, with some extra requirements:
 - Your secondary node should be allowed to communicate via HTTP/HTTPS and
   SSH with your primary node (make sure your firewall is not blocking that).
 
+### Database Encryption Key
+
+GitLab stores a unique encryption key in disk that we use to safely store sensitive
+data in the database.
+
+Any secondary node must have the exact same value for `db_key_base` as defined in the primary one.
+
+For Omnibus installations it is stored at `/etc/gitlab/gitlab-secrets.json`.
+
+For Source installations it is stored at `/home/git/gitlab/config/secrets.yml`.
+
+
 ### Authorized keys regeneration
 
 The final step will be to regenerate the keys for `.ssh/authorized_keys` using

doc/gitlab-geo/database.md

@@ -4,12 +4,13 @@ This document describes the minimal steps you have to take in order to
 replicate your GitLab database into another server. You may have to change
 some values according to your database setup, how big it is, etc.
 
-The GitLab primary node where the write operations happen will act as `master`,
-and the secondary ones which are read-only will act as `slaves`.
+The GitLab primary node where the write operations happen will connect to
+`primary` database server, and the secondary ones which are read-only will
+connect to `secondary` database servers (which are read-only too).
 
 >**Note:**
-To be on par with GitLab's notation, we will use `primary` to denote the `master`
-server, and `secondary` for the `slave`.
+In many databases documentation you will see `primary` being references as `master`
+and `secondary` as either `slave` or `standby` server (read-only).
 
 <!-- START doctoc generated TOC please keep comment here to allow auto update -->
 <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->

lib/api/geo.rb

@@ -30,6 +30,15 @@ module API
         when 'tag_push'
           required_attributes! %w(event_name project_id project)
           ::Geo::ScheduleWikiRepoUpdateService.new(params).execute
+        when 'project_destroy'
+          required_attributes! %w(event_name project_id path_with_namespace)
+          ::Geo::ScheduleRepoDestroyService.new(params).execute
+        when 'project_rename'
+          required_attributes! %w(event_name project_id path_with_namespace old_path_with_namespace)
+          ::Geo::ScheduleRepoRenameService.new(params).execute
+        when 'project_transfer'
+          required_attributes! %w(event_name project_id path_with_namespace old_path_with_namespace)
+          ::Geo::ScheduleRepoRenameService.new(params).execute
         end
       end
     end

lib/gitlab/geo.rb

@@ -22,6 +22,10 @@ module Gitlab
       RequestStore.store[:geo_node_enabled] ||= GeoNode.exists?
     end
 
+    def self.license_allows?
+      ::License.current && ::License.current.add_on?('GitLab_Geo')
+    end
+
     def self.primary?
       RequestStore.store[:geo_node_primary?] ||= self.enabled? && self.current_node && self.current_node.primary?
     end

lib/gitlab/git_access.rb

@@ -70,6 +70,10 @@ module Gitlab
         return build_status_object(false, 'The project you were looking for could not be found.')
       end
 
+      if Gitlab::Geo.secondary? && !Gitlab::Geo.license_allows?
+        return build_status_object(false, 'Your current license does not have GitLab Geo add-on enabled.')
+      end
+
       case cmd
       when *DOWNLOAD_COMMANDS
         download_access_check
@@ -94,7 +98,7 @@ module Gitlab
 
     def push_access_check(changes)
 
-      if Gitlab::Geo.enabled? && Gitlab::Geo.secondary?
+      if Gitlab::Geo.secondary?
         return build_status_object(false, "You can't push code on a secondary GitLab Geo node.")
       end
 

lib/gitlab/middleware/readonly_geo.rb

@@ -13,8 +13,8 @@ module Gitlab
         @env = env
 
         if disallowed_request? && Gitlab::Geo.secondary?
-          Rails.logger.debug('Gitlab Geo: preventing possible non readonly operation')
-          error_message = 'You cannot do writing operations on a secondary Gitlab Geo instance'
+          Rails.logger.debug('GitLab Geo: preventing possible non readonly operation')
+          error_message = 'You cannot do writing operations on a secondary GitLab Geo instance'
 
           if json_request?
             return [403, { 'Content-Type' => 'application/json' }, [{ 'message' => error_message }.to_json]]
@@ -60,12 +60,20 @@ module Gitlab
       end
 
       def whitelisted_routes
-        logout_route || WHITELISTED.any? { |path| @request.path.include?(path) }
+        logout_route || grack_route || WHITELISTED.any? { |path| @request.path.include?(path) } || sidekiq_route
       end
 
       def logout_route
         route_hash[:controller] == 'sessions' && route_hash[:action] == 'destroy'
       end
+
+      def sidekiq_route
+        @request.path.start_with?('/admin/sidekiq')
+      end
+
+      def grack_route
+        @request.path.end_with?('.git/git-upload-pack')
+      end
     end
   end
 end

spec/lib/gitlab/geo_spec.rb

@@ -68,4 +68,21 @@ describe Gitlab::Geo, lib: true do
       expect(described_class.geo_node?(host: 'inexistent', port: 1234)).to be_falsey
     end
   end
+
+  describe 'license_allows?' do
+    it 'returns true if license has Geo addon' do
+      allow_any_instance_of(License).to receive(:add_on?).with('GitLab_Geo') { true }
+      expect(described_class.license_allows?).to be_truthy
+    end
+
+    it 'returns false if license doesnt have Geo addon' do
+      allow_any_instance_of(License).to receive(:add_on?).with('GitLab_Geo') { false }
+      expect(described_class.license_allows?).to be_falsey
+    end
+
+    it 'returns false if no license is present' do
+      allow(License).to receive(:current) { nil }
+      expect(described_class.license_allows?).to be_falsey
+    end
+  end
 end