Use find_in_batches for rake task

1 files changed, 11 insertions, 3 deletions

diff --git a/lib/tasks/gitlab/traces.rake b/lib/tasks/gitlab/traces.rake
index f45edc922aa..04b939d23c9 100644
--- a/lib/tasks/gitlab/traces.rake
+++ b/lib/tasks/gitlab/traces.rake
@@ -8,10 +8,18 @@ namespace :gitlab do
       logger = Logger.new(STDOUT)
       logger.info('Archiving legacy traces')
 
-      job_ids = Ci::Build.complete.without_trace_artifact.pluck(:id)
-      job_ids = job_ids.map { |build_id| [build_id] }
+      Ci::Build.joins('RIGHT JOIN ci_job_artifacts ON ci_job_artifacts.job_id = ci_builds.id')
+        .finished
+        .where('ci_job_artifacts.file_type <> 3')
+        .group('ci_builds.id')
+        .order(id: :asc)
+        .find_in_batches(batch_size: 1000) do |jobs|
+        job_ids = jobs.map { |job| [job.id] }
 
-      ArchiveLegacyTraceWorker.bulk_perform_async(job_ids)
+        ArchiveLegacyTraceWorker.bulk_perform_async(job_ids)
+
+        logger.info("Scheduled #{job_ids.count} jobs. From #{job_ids.min} #{job_ids.max}")
+      end
     end
   end
 end