[Findit] Fixing when try job entities are created

appengine/findit/waterfall/try_job_util.py

Index: appengine/findit/waterfall/try_job_util.py
diff --git a/appengine/findit/waterfall/try_job_util.py b/appengine/findit/waterfall/try_job_util.py
index 0ea4d52f0279bde54554a273b12e89e6babbbd43..3aa160aead843e067ba1783928cb4c05036f7386 100644
--- a/appengine/findit/waterfall/try_job_util.py
+++ b/appengine/findit/waterfall/try_job_util.py
@@ -97,8 +97,7 @@ def _NeedANewTryJob(
             failed_steps))
 
   if need_new_try_job:
-    try_job = WfTryJob.Get(
-        master_name, builder_name, build_number)
+    try_job = WfTryJob.Get(master_name, builder_name, build_number)
 
     if try_job:
       if try_job.failed:
@@ -106,10 +105,6 @@ def _NeedANewTryJob(
         try_job.put()
       else:
         need_new_try_job = False
-    else:
-      try_job = WfTryJob.Create(
-          master_name, builder_name, build_number)
-      try_job.put()
 
   return need_new_try_job, last_pass, try_job_type, targeted_tests
 
@@ -169,6 +164,9 @@ def ScheduleTryJobIfNeeded(failure_info, signals, heuristic_result):
     return {}
 
   if need_new_try_job:
+    try_job = WfTryJob.Create(master_name, builder_name, build_number)

[stgao, 2016/04/18 22:05:27]

No, we'd better not do it like this.
If there are two pipeline running in parallel, we might trigger two try-jobs
with this change.

I thought I have added the failure type to ``failure_info``, can we use that for
the checking above?

+    try_job.put()
+
     compile_targets = (_GetFailedTargetsFromSignals(
         signals, master_name, builder_name)
                        if try_job_type == TryJobType.COMPILE else None)