[Findit] Report the error from the last flake swarming task that failed in master_flake_analysis.py

appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py

Index: appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py
diff --git a/appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py b/appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py
index 881aef8c9dc6270df9cd995266c3bf3412a0e330..c8d32f52c0690a44c53e2b6f0b2a28c81a6cf51b 100644
--- a/appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py
+++ b/appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py
@@ -866,6 +866,17 @@ class RecursiveFlakePipelineTest(wf_testcase.WaterfallTestCase):
         analysis, analysis_status.COMPLETED, None)
     self.assertEqual(analysis.result_status, result_status.FOUND_UNTRIAGED)
 
+  def testUpdateAnalysisUponCompletionError(self):

[chanli, 2016/11/28 22:27:54]

This test case is not what I expect for your new change? Shouldn't you have a
test case where you have a failed flake_swarming_task and use the error message
from the task?

[lijeffrey, 2016/11/28 23:37:09]

Done.

+    expected_error = {
+        'code': 1,
+        'message': 'some error message'
+    }
+    analysis = MasterFlakeAnalysis.Create('m', 'b', 123, 's', 't')
+    analysis.suspected_flake_build_number = 100
+    recursive_flake_pipeline._UpdateAnalysisStatusUponCompletion(
+        analysis, analysis_status.COMPLETED, expected_error)
+    self.assertEqual(expected_error, analysis.error)
+
   def testGetListOfNearbyBuildNumbers(self):
     self.assertEqual(
         [1],