Rewrite the GitHub importer from scratch

Prior to this MR there were two GitHub related importers:

* Github::Import: the main importer used for GitHub projects
* Gitlab::GithubImport: importer that's somewhat confusingly used for
  importing Gitea projects (apparently they have a compatible API)

This MR renames the Gitea importer to Gitlab::LegacyGithubImport and
introduces a new GitHub importer in the Gitlab::GithubImport namespace.
This new GitHub importer uses Sidekiq for importing multiple resources
in parallel, though it also has the ability to import data sequentially
should this be necessary.

The new code is spread across the following directories:

* lib/gitlab/github_import: this directory contains most of the importer
  code such as the classes used for importing resources.
* app/workers/gitlab/github_import: this directory contains the Sidekiq
  workers, most of which simply use the code from the directory above.
* app/workers/concerns/gitlab/github_import: this directory provides a
  few modules that are included in every GitHub importer worker.

== Stages

The import work is divided into separate stages, with each stage
importing a specific set of data. Stages will schedule the work that
needs to be performed, followed by scheduling a job for the
"AdvanceStageWorker" worker. This worker will periodically check if all
work is completed and schedule the next stage if this is the case. If
work is not yet completed this worker will reschedule itself.

Using this approach we don't have to block threads by calling `sleep()`,
as doing so for large projects could block the thread from doing any
work for many hours.

== Retrying Work

Workers will reschedule themselves whenever necessary. For example,
hitting the GitHub API's rate limit will result in jobs rescheduling
themselves. These jobs are not processed until the rate limit has been
reset.

== User Lookups

Part of the importing process involves looking up user details in the
GitHub API so we can map them to GitLab users. The old importer used
an in-memory cache, but this obviously doesn't work when the work is
spread across different threads.

The new importer uses a Redis cache and makes sure we only perform
API/database calls if absolutely necessary.  Frequently used keys are
refreshed, and lookup misses are also cached; removing the need for
performing API/database calls if we know we don't have the data we're
looking for.

== Performance & Models

The new importer in various places uses raw INSERT statements (as
generated by `Gitlab::Database.bulk_insert`) instead of using Rails
models. This allows us to bypass any validations and callbacks,
drastically reducing the number of SQL queries and Gitaly RPC calls
necessary to import projects.

To ensure the code produces valid data the corresponding tests check if
the produced rows are valid according to the model validation rules.

18 files changed, 592 insertions, 3 deletions

diff --git a/app/workers/concerns/gitlab/github_import/notify_upon_death.rb b/app/workers/concerns/gitlab/github_import/notify_upon_death.rb
new file mode 100644
index 00000000000..3d7120665b6
--- /dev/null
+++ b/app/workers/concerns/gitlab/github_import/notify_upon_death.rb
@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    # NotifyUponDeath can be included into a GitHub worker class if it should
+    # notify any JobWaiter instances upon being moved to the Sidekiq dead queue.
+    #
+    # Note that this will only notify the waiter upon graceful termination, a
+    # SIGKILL will still result in the waiter _not_ being notified.
+    #
+    # Workers including this module must have jobs passed where the last
+    # argument is the key to notify, as a String.
+    module NotifyUponDeath
+      extend ActiveSupport::Concern
+
+      included do
+        # If a job is being exhausted we still want to notify the
+        # AdvanceStageWorker. This prevents the entire import from getting stuck
+        # just because 1 job threw too many errors.
+        sidekiq_retries_exhausted do |job|
+          args = job['args']
+          jid = job['jid']
+
+          if args.length == 3 && (key = args.last) && key.is_a?(String)
+            JobWaiter.notify(key, jid)
+          end
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/concerns/gitlab/github_import/object_importer.rb b/app/workers/concerns/gitlab/github_import/object_importer.rb
new file mode 100644
index 00000000000..67e36c811de
--- /dev/null
+++ b/app/workers/concerns/gitlab/github_import/object_importer.rb
@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    # ObjectImporter defines the base behaviour for every Sidekiq worker that
+    # imports a single resource such as a note or pull request.
+    module ObjectImporter
+      extend ActiveSupport::Concern
+
+      included do
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include ReschedulingMethods
+        include NotifyUponDeath
+      end
+
+      # project - An instance of `Project` to import the data into.
+      # client - An instance of `Gitlab::GithubImport::Client`
+      # hash - A Hash containing the details of the object to import.
+      def import(project, client, hash)
+        object = representation_class.from_json_hash(hash)
+
+        importer_class.new(object, project, client).execute
+
+        counter.increment(project: project.path_with_namespace)
+      end
+
+      def counter
+        @counter ||= Gitlab::Metrics.counter(counter_name, counter_description)
+      end
+
+      # Returns the representation class to use for the object. This class must
+      # define the class method `from_json_hash`.
+      def representation_class
+        raise NotImplementedError
+      end
+
+      # Returns the class to use for importing the object.
+      def importer_class
+        raise NotImplementedError
+      end
+
+      # Returns the name (as a Symbol) of the Prometheus counter.
+      def counter_name
+        raise NotImplementedError
+      end
+
+      # Returns the description (as a String) of the Prometheus counter.
+      def counter_description
+        raise NotImplementedError
+      end
+    end
+  end
+end
diff --git a/app/workers/concerns/gitlab/github_import/queue.rb b/app/workers/concerns/gitlab/github_import/queue.rb
new file mode 100644
index 00000000000..a2bee361b86
--- /dev/null
+++ b/app/workers/concerns/gitlab/github_import/queue.rb
@@ -0,0 +1,16 @@
+module Gitlab
+  module GithubImport
+    module Queue
+      extend ActiveSupport::Concern
+
+      included do
+        # If a job produces an error it may block a stage from advancing
+        # forever. To prevent this from happening we prevent jobs from going to
+        # the dead queue. This does mean some resources may not be imported, but
+        # this is better than a project being stuck in the "import" state
+        # forever.
+        sidekiq_options queue: 'github_importer', dead: false, retry: 5
+      end
+    end
+  end
+end
diff --git a/app/workers/concerns/gitlab/github_import/rescheduling_methods.rb b/app/workers/concerns/gitlab/github_import/rescheduling_methods.rb
new file mode 100644
index 00000000000..692ca6b7f42
--- /dev/null
+++ b/app/workers/concerns/gitlab/github_import/rescheduling_methods.rb
@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    # Module that provides methods shared by the various workers used for
+    # importing GitHub projects.
+    module ReschedulingMethods
+      # project_id - The ID of the GitLab project to import the note into.
+      # hash - A Hash containing the details of the GitHub object to imoprt.
+      # notify_key - The Redis key to notify upon completion, if any.
+      def perform(project_id, hash, notify_key = nil)
+        project = Project.find_by(id: project_id)
+
+        return notify_waiter(notify_key) unless project
+
+        client = GithubImport.new_client_for(project, parallel: true)
+
+        if try_import(project, client, hash)
+          notify_waiter(notify_key)
+        else
+          # In the event of hitting the rate limit we want to reschedule the job
+          # so its retried after our rate limit has been reset.
+          self.class
+            .perform_in(client.rate_limit_resets_in, project.id, hash, notify_key)
+        end
+      end
+
+      def try_import(*args)
+        import(*args)
+        true
+      rescue RateLimitError
+        false
+      end
+
+      def notify_waiter(key = nil)
+        JobWaiter.notify(key, jid) if key
+      end
+    end
+  end
+end
diff --git a/app/workers/concerns/gitlab/github_import/stage_methods.rb b/app/workers/concerns/gitlab/github_import/stage_methods.rb
new file mode 100644
index 00000000000..147c8c8d683
--- /dev/null
+++ b/app/workers/concerns/gitlab/github_import/stage_methods.rb
@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module StageMethods
+      # project_id - The ID of the GitLab project to import the data into.
+      def perform(project_id)
+        return unless (project = find_project(project_id))
+
+        client = GithubImport.new_client_for(project)
+
+        try_import(client, project)
+      end
+
+      # client - An instance of Gitlab::GithubImport::Client.
+      # project - An instance of Project.
+      def try_import(client, project)
+        import(client, project)
+      rescue RateLimitError
+        self.class.perform_in(client.rate_limit_resets_in, project.id)
+      end
+
+      def find_project(id)
+        # If the project has been marked as failed we want to bail out
+        # automatically.
+        Project.import_started.find_by(id: id)
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/advance_stage_worker.rb b/app/workers/gitlab/github_import/advance_stage_worker.rb
new file mode 100644
index 00000000000..877f88c043f
--- /dev/null
+++ b/app/workers/gitlab/github_import/advance_stage_worker.rb
@@ -0,0 +1,74 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    # AdvanceStageWorker is a worker used by the GitHub importer to wait for a
+    # number of jobs to complete, without blocking a thread. Once all jobs have
+    # been completed this worker will advance the import process to the next
+    # stage.
+    class AdvanceStageWorker
+      include Sidekiq::Worker
+
+      sidekiq_options queue: 'github_importer_advance_stage', dead: false
+
+      INTERVAL = 30.seconds.to_i
+
+      # The number of seconds to wait (while blocking the thread) before
+      # continueing to the next waiter.
+      BLOCKING_WAIT_TIME = 5
+
+      # The known importer stages and their corresponding Sidekiq workers.
+      STAGES = {
+        issues_and_diff_notes: Stage::ImportIssuesAndDiffNotesWorker,
+        notes: Stage::ImportNotesWorker,
+        finish: Stage::FinishImportWorker
+      }.freeze
+
+      # project_id - The ID of the project being imported.
+      # waiters - A Hash mapping Gitlab::JobWaiter keys to the number of
+      #           remaining jobs.
+      # next_stage - The name of the next stage to start when all jobs have been
+      #              completed.
+      def perform(project_id, waiters, next_stage)
+        return unless (project = find_project(project_id))
+
+        new_waiters = wait_for_jobs(waiters)
+
+        if new_waiters.empty?
+          # We refresh the import JID here so workers importing individual
+          # resources (e.g. notes) don't have to do this all the time, reducing
+          # the pressure on Redis. We _only_ do this once all jobs are done so
+          # we don't get stuck forever if one or more jobs failed to notify the
+          # JobWaiter.
+          project.refresh_import_jid_expiration
+
+          STAGES.fetch(next_stage.to_sym).perform_async(project_id)
+        else
+          self.class.perform_in(INTERVAL, project_id, new_waiters, next_stage)
+        end
+      end
+
+      def wait_for_jobs(waiters)
+        waiters.each_with_object({}) do |(key, remaining), new_waiters|
+          waiter = JobWaiter.new(remaining, key)
+
+          # We wait for a brief moment of time so we don't reschedule if we can
+          # complete the work fast enough.
+          waiter.wait(BLOCKING_WAIT_TIME)
+
+          next unless waiter.jobs_remaining.positive?
+
+          new_waiters[waiter.key] = waiter.jobs_remaining
+        end
+      end
+
+      def find_project(id)
+        # We only care about the import JID so we can refresh it. We also only
+        # want the project if it hasn't been marked as failed yet. It's possible
+        # the import gets marked as stuck when jobs of the current stage failed
+        # somehow.
+        Project.select(:import_jid).import_started.find_by(id: id)
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/import_diff_note_worker.rb b/app/workers/gitlab/github_import/import_diff_note_worker.rb
new file mode 100644
index 00000000000..ef2a74c51c5
--- /dev/null
+++ b/app/workers/gitlab/github_import/import_diff_note_worker.rb
@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    class ImportDiffNoteWorker
+      include ObjectImporter
+
+      def representation_class
+        Representation::DiffNote
+      end
+
+      def importer_class
+        Importer::DiffNoteImporter
+      end
+
+      def counter_name
+        :github_importer_imported_diff_notes
+      end
+
+      def counter_description
+        'The number of imported GitHub pull request review comments'
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/import_issue_worker.rb b/app/workers/gitlab/github_import/import_issue_worker.rb
new file mode 100644
index 00000000000..1b081ae5966
--- /dev/null
+++ b/app/workers/gitlab/github_import/import_issue_worker.rb
@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    class ImportIssueWorker
+      include ObjectImporter
+
+      def representation_class
+        Representation::Issue
+      end
+
+      def importer_class
+        Importer::IssueAndLabelLinksImporter
+      end
+
+      def counter_name
+        :github_importer_imported_issues
+      end
+
+      def counter_description
+        'The number of imported GitHub issues'
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/import_note_worker.rb b/app/workers/gitlab/github_import/import_note_worker.rb
new file mode 100644
index 00000000000..d2b4c36a5b9
--- /dev/null
+++ b/app/workers/gitlab/github_import/import_note_worker.rb
@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    class ImportNoteWorker
+      include ObjectImporter
+
+      def representation_class
+        Representation::Note
+      end
+
+      def importer_class
+        Importer::NoteImporter
+      end
+
+      def counter_name
+        :github_importer_imported_notes
+      end
+
+      def counter_description
+        'The number of imported GitHub comments'
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/import_pull_request_worker.rb b/app/workers/gitlab/github_import/import_pull_request_worker.rb
new file mode 100644
index 00000000000..62a6da152a3
--- /dev/null
+++ b/app/workers/gitlab/github_import/import_pull_request_worker.rb
@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    class ImportPullRequestWorker
+      include ObjectImporter
+
+      def representation_class
+        Representation::PullRequest
+      end
+
+      def importer_class
+        Importer::PullRequestImporter
+      end
+
+      def counter_name
+        :github_importer_imported_pull_requests
+      end
+
+      def counter_description
+        'The number of imported GitHub pull requests'
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/refresh_import_jid_worker.rb b/app/workers/gitlab/github_import/refresh_import_jid_worker.rb
new file mode 100644
index 00000000000..45a38927225
--- /dev/null
+++ b/app/workers/gitlab/github_import/refresh_import_jid_worker.rb
@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    class RefreshImportJidWorker
+      include Sidekiq::Worker
+      include GithubImport::Queue
+
+      # The interval to schedule new instances of this job at.
+      INTERVAL = 1.minute.to_i
+
+      def self.perform_in_the_future(*args)
+        perform_in(INTERVAL, *args)
+      end
+
+      # project_id - The ID of the project that is being imported.
+      # check_job_id - The ID of the job for which to check the status.
+      def perform(project_id, check_job_id)
+        return unless (project = find_project(project_id))
+
+        if SidekiqStatus.running?(check_job_id)
+          # As long as the repository is being cloned we want to keep refreshing
+          # the import JID status.
+          project.refresh_import_jid_expiration
+          self.class.perform_in_the_future(project_id, check_job_id)
+        end
+
+        # If the job is no longer running there's nothing else we need to do. If
+        # the clone job completed successfully it will have scheduled the next
+        # stage, if it died there's nothing we can do anyway.
+      end
+
+      def find_project(id)
+        Project.select(:import_jid).import_started.find_by(id: id)
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/finish_import_worker.rb b/app/workers/gitlab/github_import/stage/finish_import_worker.rb
new file mode 100644
index 00000000000..1a09497780a
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/finish_import_worker.rb
@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class FinishImportWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # project - An instance of Project.
+        def import(_, project)
+          project.after_import
+          report_import_time(project)
+        end
+
+        def report_import_time(project)
+          duration = Time.zone.now - project.created_at
+          path = project.path_with_namespace
+
+          histogram.observe({ project: path }, duration)
+          counter.increment
+
+          logger.info("GitHub importer finished for #{path} in #{duration.round(2)} seconds")
+        end
+
+        def histogram
+          @histogram ||= Gitlab::Metrics.histogram(
+            :github_importer_total_duration_seconds,
+            'Total time spent importing GitHub projects, in seconds'
+          )
+        end
+
+        def counter
+          @counter ||= Gitlab::Metrics.counter(
+            :github_importer_imported_projects,
+            'The number of imported GitHub projects'
+          )
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/import_base_data_worker.rb b/app/workers/gitlab/github_import/stage/import_base_data_worker.rb
new file mode 100644
index 00000000000..f8a3684c6ba
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/import_base_data_worker.rb
@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class ImportBaseDataWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # These importers are fast enough that we can just run them in the same
+        # thread.
+        IMPORTERS = [
+          Importer::LabelsImporter,
+          Importer::MilestonesImporter,
+          Importer::ReleasesImporter
+        ].freeze
+
+        # client - An instance of Gitlab::GithubImport::Client.
+        # project - An instance of Project.
+        def import(client, project)
+          IMPORTERS.each do |klass|
+            klass.new(project, client).execute
+          end
+
+          project.refresh_import_jid_expiration
+
+          ImportPullRequestsWorker.perform_async(project.id)
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/import_issues_and_diff_notes_worker.rb b/app/workers/gitlab/github_import/stage/import_issues_and_diff_notes_worker.rb
new file mode 100644
index 00000000000..e110b7c1c36
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/import_issues_and_diff_notes_worker.rb
@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class ImportIssuesAndDiffNotesWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # The importers to run in this stage. Issues can't be imported earlier
+        # on as we also use these to enrich pull requests with assigned labels.
+        IMPORTERS = [
+          Importer::IssuesImporter,
+          Importer::DiffNotesImporter
+        ].freeze
+
+        # client - An instance of Gitlab::GithubImport::Client.
+        # project - An instance of Project.
+        def import(client, project)
+          waiters = IMPORTERS.each_with_object({}) do |klass, hash|
+            waiter = klass.new(project, client).execute
+            hash[waiter.key] = waiter.jobs_remaining
+          end
+
+          AdvanceStageWorker.perform_async(project.id, waiters, :notes)
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/import_notes_worker.rb b/app/workers/gitlab/github_import/stage/import_notes_worker.rb
new file mode 100644
index 00000000000..9810ed25cf9
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/import_notes_worker.rb
@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class ImportNotesWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # client - An instance of Gitlab::GithubImport::Client.
+        # project - An instance of Project.
+        def import(client, project)
+          waiter = Importer::NotesImporter
+            .new(project, client)
+            .execute
+
+          AdvanceStageWorker.perform_async(
+            project.id,
+            { waiter.key => waiter.jobs_remaining },
+            :finish
+          )
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/import_pull_requests_worker.rb b/app/workers/gitlab/github_import/stage/import_pull_requests_worker.rb
new file mode 100644
index 00000000000..c531f26e897
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/import_pull_requests_worker.rb
@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class ImportPullRequestsWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # client - An instance of Gitlab::GithubImport::Client.
+        # project - An instance of Project.
+        def import(client, project)
+          waiter = Importer::PullRequestsImporter
+            .new(project, client)
+            .execute
+
+          project.refresh_import_jid_expiration
+
+          AdvanceStageWorker.perform_async(
+            project.id,
+            { waiter.key => waiter.jobs_remaining },
+            :issues_and_diff_notes
+          )
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/gitlab/github_import/stage/import_repository_worker.rb b/app/workers/gitlab/github_import/stage/import_repository_worker.rb
new file mode 100644
index 00000000000..aa5762e773d
--- /dev/null
+++ b/app/workers/gitlab/github_import/stage/import_repository_worker.rb
@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module GithubImport
+    module Stage
+      class ImportRepositoryWorker
+        include Sidekiq::Worker
+        include GithubImport::Queue
+        include StageMethods
+
+        # client - An instance of Gitlab::GithubImport::Client.
+        # project - An instance of Project.
+        def import(client, project)
+          # In extreme cases it's possible for a clone to take more than the
+          # import job expiration time. To work around this we schedule a
+          # separate job that will periodically run and refresh the import
+          # expiration time.
+          RefreshImportJidWorker.perform_in_the_future(project.id, jid)
+
+          importer = Importer::RepositoryImporter.new(project, client)
+
+          return unless importer.execute
+
+          counter.increment
+
+          ImportBaseDataWorker.perform_async(project.id)
+        end
+
+        def counter
+          Gitlab::Metrics.counter(
+            :github_importer_imported_repositories,
+            'The number of imported GitHub repositories'
+          )
+        end
+      end
+    end
+  end
+end
diff --git a/app/workers/repository_import_worker.rb b/app/workers/repository_import_worker.rb
index d7c0043d3b6..4e90b137b26 100644
--- a/app/workers/repository_import_worker.rb
+++ b/app/workers/repository_import_worker.rb
@@ -17,11 +17,16 @@ class RepositoryImportWorker
                               import_url: project.import_url,
                               path: project.full_path)
 
-    result = Projects::ImportService.new(project, project.creator).execute
+    service = Projects::ImportService.new(project, project.creator)
+    result = service.execute
+
+    # Some importers may perform their work asynchronously. In this case it's up
+    # to those importers to mark the import process as complete.
+    return if service.async?
+
     raise ImportError, result[:message] if result[:status] == :error
 
-    project.repository.after_import
-    project.import_finish
+    project.after_import
   rescue ImportError => ex
     fail_import(project, ex.message)
     raise