1 files changed, 34 insertions, 1 deletions
diff --git a/app/models/ci/build_trace_chunks/fog.rb b/app/models/ci/build_trace_chunks/fog.rb
index fab85fae33d..3bfac2b33c0 100644
--- a/app/models/ci/build_trace_chunks/fog.rb
+++ b/app/models/ci/build_trace_chunks/fog.rb
@@ -25,14 +25,36 @@ module Ci
         files.create(create_attributes(model, new_data))
       end
 
+      # This is the sequence that causes append_data to be called:
+      #
+      # 1. Runner sends a PUT /api/v4/jobs/:id to indicate the job is canceled or finished.
+      # 2. UpdateBuildStateService#accept_build_state! persists all live job logs to object storage (or filesystem).
+      # 3. UpdateBuildStateService#accept_build_state! returns a 202 to the runner.
+      # 4. The runner continues to send PATCH requests with job logs until all logs have been sent and received.
+      # 5. If the last PATCH request arrives after the job log has been persisted, we
+      #    retrieve the data from object storage to append the remaining lines.
       def append_data(model, new_data, offset)
         if offset > 0
           truncated_data = data(model).to_s.byteslice(0, offset)
-          new_data = truncated_data + new_data
+          new_data = append_strings(truncated_data, new_data)
         end
 
         set_data(model, new_data)
         new_data.bytesize
+      rescue Encoding::CompatibilityError => e
+        Gitlab::ErrorTracking.track_and_raise_exception(
+          e,
+          build_id: model.build_id,
+          chunk_index: model.chunk_index,
+          chunk_start_offset: model.start_offset,
+          chunk_end_offset: model.end_offset,
+          chunk_size: model.size,
+          chunk_data_store: model.data_store,
+          offset: offset,
+          old_data_encoding: truncated_data.encoding.to_s,
+          new_data: new_data,
+          new_data_size: new_data.bytesize,
+          new_data_encoding: new_data.encoding.to_s)
       end
 
       def size(model)
@@ -57,6 +79,17 @@ module Ci
 
       private
 
+      def append_strings(old_data, new_data)
+        if Feature.enabled?(:ci_job_trace_force_encode, default_enabled: :yaml)
+          # When object storage is in use, old_data may be retrieved in UTF-8.
+          old_data = old_data.force_encoding(Encoding::ASCII_8BIT)
+          # new_data should already be in ASCII-8BIT, but just in case it isn't, do this.
+          new_data = new_data.force_encoding(Encoding::ASCII_8BIT)
+        end
+
+        old_data + new_data
+      end
+
       def key(model)
         key_raw(model.build_id, model.chunk_index)
       end