# frozen_string_literal: true module Gitlab module BitbucketServerImport module ParallelScheduling include Loggable attr_reader :project, :page_counter, :already_processed_cache_key, :job_waiter_cache_key, :job_waiter_remaining_cache_key attr_accessor :enqueued_job_counter # The base cache key to use for tracking already processed objects. ALREADY_PROCESSED_CACHE_KEY = 'bitbucket-server-importer/already-processed/%{project}/%{collection}' # The base cache key to use for storing job waiter key JOB_WAITER_CACHE_KEY = 'bitbucket-server-importer/job-waiter/%{project}/%{collection}' # The base cache key to use for storing job waiter remaining jobs JOB_WAITER_REMAINING_CACHE_KEY = 'bitbucket-server-importer/job-waiter-remaining/%{project}/%{collection}' # project - An instance of `Project`. def initialize(project) @project = project @page_counter = Gitlab::Import::PageCounter.new(project, collection_method, 'bitbucket-server-importer') @already_processed_cache_key = format(ALREADY_PROCESSED_CACHE_KEY, project: project.id, collection: collection_method) @job_waiter_cache_key = format(JOB_WAITER_CACHE_KEY, project: project.id, collection: collection_method) @job_waiter_remaining_cache_key = format(JOB_WAITER_REMAINING_CACHE_KEY, project: project.id, collection: collection_method) # The enqueued job counter is used to calculate job delays and distribute # them over time. When the stage worker restarts, the counter resets to # prevent jobs from being queued too far into the future. Such logic may # result in more jobs being executed when the stage worker resumes. An # alternative solution would complicate the delay logic, so for simplicity # we accept more jobs being executed. @enqueued_job_counter = 0 end private def client @client ||= BitbucketServer::Client.new(project.import_data.credentials) end def project_key @project_key ||= project.import_data.data['project_key'] end def repository_slug @repository_slug ||= project.import_data.data['repo_slug'] end # Returns the ID to use for the cache used for checking if an object has # already been processed or not. # # object - The object we may want to import. def id_for_already_processed_cache(object) raise NotImplementedError end # The Sidekiq worker class used for scheduling the importing of objects in # parallel. def sidekiq_worker_class raise NotImplementedError end # The name of the method to call to retrieve the data to import. def collection_method raise NotImplementedError end def job_waiter @job_waiter ||= begin key = Gitlab::Cache::Import::Caching.read(job_waiter_cache_key) key ||= Gitlab::Cache::Import::Caching.write(job_waiter_cache_key, JobWaiter.generate_key) jobs_remaining = Gitlab::Cache::Import::Caching.read(job_waiter_remaining_cache_key).to_i || 0 JobWaiter.new(jobs_remaining, key) end end def already_processed?(object) id = id_for_already_processed_cache(object) Gitlab::Cache::Import::Caching.set_includes?(already_processed_cache_key, id) end # Marks the given object as "already processed". def mark_as_processed(object) id = id_for_already_processed_cache(object) Gitlab::Cache::Import::Caching.set_add(already_processed_cache_key, id) end def calculate_job_delay(job_index) runtime = Time.current - job_started_at multiplier = (job_index / concurrent_import_jobs_limit.to_f) (multiplier * 1.minute) + 1.second - runtime end def job_started_at @job_started_at ||= Time.current end def concurrent_import_jobs_limit Gitlab::CurrentSettings.concurrent_bitbucket_server_import_jobs_limit end def track_import_failure!(project, exception:, **args) Gitlab::Import::ImportFailureService.track( project_id: project.id, error_source: self.class.name, exception: exception, **args ) end end end end