Issue 796438  has been merged into this issue.

OK.  I did some digging, and then some more digging.

The suite job is here:
    http://cautotest-prod.corp.google.com/afe/#tab_id=view_job&object_id=163820593

That shows every test completing successfully, except for one aborted
job:
    http://cautotest-prod.corp.google.com/afe/#tab_id=view_job&object_id=163820700

If you look in the suite job logs, you see that this job isn't
listed as finishing, and that every other job completed by 15:12:02 at
the latest.  Between that point and suite abort at 16:31:29, there
was a whole lot of idle DUTs, and no testing.

Going to the shard, and looking in the scheduler logs, you see all
of the following entries:
12/19 15:02:38.441 INFO |  scheduler_models:0588| HQE: 164207534, for job: 163820700 and host: chromeos4-row5-rack9-host21 has status:Starting [active] -> Running
12/19 15:02:38.452 INFO |         rdb_hosts:0222| Host chromeos4-row5-rack9-host21 in Pending updating {'status': 'Running'} through rdb on behalf of: HQE: 164207534, for job: 163820700 
12/19 15:02:38.458 INFO |         rdb_hosts:0222| Host chromeos4-row5-rack9-host21 in Running updating {'dirty': 1} through rdb on behalf of: HQE: 164207534, for job: 163820700 
12/19 15:03:12.504 INFO |  scheduler_models:0588| HQE: 164207534, for job: 163820700 and host: chromeos4-row5-rack9-host21 has status:Running [active] -> Gathering
12/19 15:03:12.508 INFO |         rdb_hosts:0222| Host chromeos4-row5-rack9-host21 in Running updating {'status': 'Running'} through rdb on behalf of: HQE: 164207534, for job: 163820700 
12/19 15:03:13.021 INFO |  scheduler_models:0588| HQE: 164207534, for job: 163820700 and host: chromeos4-row5-rack9-host21 has status:Gathering [active] -> Parsing
12/19 15:03:13.027 INFO |         rdb_hosts:0222| Host chromeos4-row5-rack9-host21 in Running updating {'status': 'Ready'} through rdb on behalf of: HQE: 164207534, for job: 163820700 

So... The job ran.  To completion.  At first blush, without incident.

And the coup de grace, on the shard, you can see this:
    chromeos-test@chromeos-server130:/usr/local/autotest/logs$ ls -d /usr/local/autotest/results/163820700*
    /usr/local/autotest/results/163820700-chromeos-test

My top theory is that the job wasn't properly transferred from the
shard to the master.  But I am only an egg...