[Findit] Implementing swarming task error detection

appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py

Index: appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py
diff --git a/appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py b/appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py
index 720cb7784c5404ee2ab4a546cdfc6911a8e0a1a7..5153a31a843f3f96fd1035da99150964da1d66f9 100644
--- a/appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py
+++ b/appengine/findit/waterfall/trigger_base_swarming_task_pipeline.py
@@ -168,11 +168,13 @@ class TriggerBaseSwarmingTaskPipeline(BasePipeline):  # pragma: no cover.
         step_name, tests, iterations_to_rerun)
 
     # 3. Trigger a new Swarming task to re-run the failed tests.
-    task_id = swarming_util.TriggerSwarmingTask(new_request, http_client)
+    task_id, error = swarming_util.TriggerSwarmingTask(new_request, http_client)
 
     # Save the task id.
     swarming_task = self._GetSwarmingTask(*call_args)
     swarming_task.task_id = task_id

[chanli, 2016/11/11 00:05:13]

If the swarming_task is not even triggered, should we save the None task_id and
parameters of the request?

[lijeffrey, 2016/11/11 20:55:41]

I think we can still save the parameters, but they wouldn't get used. This can
be a design choice, I can't think of a clear reason why or why not to go either
way. Maybe in case we want to implement retry with the same parameters as before
that failed, but I don't feel strongly about going either way

+    if error:
+      swarming_task.error = error
     swarming_task.parameters['tests'] = tests
     swarming_task.parameters['iterations_to_rerun'] = iterations_to_rerun
     swarming_task.parameters['ref_name'] = swarming_util.GetTagValue(