Same for the below:
https://build.chromium.org/p/chromium.win/builders/Win8%20GN%20(dbg)/builds/23939
https://build.chromium.org/p/chromium.win/builders/Win%20x64%20Builder%20(dbg)/builds/21303

The try-job result was recorded and these builds show up on the dashboard as "Untriaged - Found".
There should be a bug.

Looking at the logs for the three builds, pipelines executed and succeeded as expected. That means WfTryJob, WfTryJobData, and WfAnalysis were updated with try-job results with the put api. But the data were not flushed into the datastore. We haven't run into such cases before.

As a remedy, I will make the update in a NDB transaction to force the data be flushed into the datastore.

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra.git/+/00689794b1ed85b90c017c2a12a0a67a04d7daa0

commit 00689794b1ed85b90c017c2a12a0a67a04d7daa0
Author: stgao <stgao@chromium.org>
Date: Mon May 02 20:49:49 2016

[Findit] Update try-job result in a transaction and fix bugs in unittests.

BUG= 607956 

Review-Url: https://codereview.chromium.org/1935193002

[modify] https://crrev.com/00689794b1ed85b90c017c2a12a0a67a04d7daa0/appengine/findit/waterfall/identify_try_job_culprit_pipeline.py
[modify] https://crrev.com/00689794b1ed85b90c017c2a12a0a67a04d7daa0/appengine/findit/waterfall/monitor_try_job_pipeline.py
[modify] https://crrev.com/00689794b1ed85b90c017c2a12a0a67a04d7daa0/appengine/findit/waterfall/test/identify_try_job_culprit_pipeline_test.py
[modify] https://crrev.com/00689794b1ed85b90c017c2a12a0a67a04d7daa0/appengine/findit/waterfall/test/swarming_tasks_to_try_job_pipeline_test.py
[modify] https://crrev.com/00689794b1ed85b90c017c2a12a0a67a04d7daa0/appengine/findit/waterfall/test/try_job_pipeline_test.py