Raise an error so that timeouts can be tracked via Sentry

app/workers/chat_notification_worker.rb

@@ -3,6 +3,9 @@
 class ChatNotificationWorker
   include ApplicationWorker
 
+  TimeoutExceeded = Class.new(StandardError)
+
+  sidekiq_options retry: false
   feature_category :chatops
   latency_sensitive_worker!
   # TODO: break this into multiple jobs
@@ -19,7 +22,7 @@ class ChatNotificationWorker
       send_response(build)
     end
   rescue Gitlab::Chat::Output::MissingBuildSectionError
-    return if timeout_exceeded?(reschedule_count)
+    raise TimeoutExceeded if timeout_exceeded?(reschedule_count)
 
     # The creation of traces and sections appears to be eventually consistent.
     # As a result it's possible for us to run the above code before the trace

spec/workers/chat_notification_worker_spec.rb

@@ -8,6 +8,10 @@ describe ChatNotificationWorker do
     create(:ci_build, pipeline: create(:ci_pipeline, source: :chat))
   end
 
+  it 'instructs sidekiq not to retry on failure' do
+    expect(described_class.get_sidekiq_options['retry']).to eq(false)
+  end
+
   describe '#perform' do
     it 'does nothing when the build no longer exists' do
       expect(worker).not_to receive(:send_response)
@@ -36,13 +40,13 @@ describe ChatNotificationWorker do
         worker.perform(chat_build.id)
       end
 
-      it "stops rescheduling the job after #{described_class::RESCHEDULE_TIMEOUT} seconds" do
+      it "raises an error after #{described_class::RESCHEDULE_TIMEOUT} seconds of retrying" do
         allow(described_class).to receive(:new).and_return(worker)
         allow(worker).to receive(:send_response).and_raise(Gitlab::Chat::Output::MissingBuildSectionError)
 
         worker.perform(chat_build.id)
 
-        described_class.drain
+        expect { described_class.drain }.to raise_error(described_class::TimeoutExceeded)
 
         max_reschedules = described_class::RESCHEDULE_TIMEOUT / described_class::RESCHEDULE_INTERVAL