This is related to bug 670000.  Server chromeos-server59.hot
was configured as a shard, but then removed from production.
The system was still trying to run shard_client.  The result
was a failure that spammed the system with e-mail like this:

Subject: monitor_db exception
chromeos-server59.hot.corp.google.com / 15689 / 09:40:27 12/05/16
EXCEPTION: Uncaught exception. Terminating shard_client.
Traceback (most recent call last):
 File "/usr/local/autotest/scheduler/shard/shard_client.py", line 386, in main
   main_without_exception_handling()
 File "/usr/local/autotest/scheduler/shard/shard_client.py", line
411, in main_without_exception_handling
   _heartbeat_client = get_shard_client()
 File "/usr/local/autotest/scheduler/shard/shard_client.py", line
378, in get_shard_client
   shard_hostname = _get_shard_hostname_and_ensure_running_on_shard()
 File "/usr/local/autotest/scheduler/shard/shard_client.py", line
359, in _get_shard_hostname_and_ensure_running_on_shard
   'To run the shard client, shard_hostname must neither be None nor '
HeartbeatOnlyAllowedInShardModeException: To run the shard client,
shard_hostname must neither be None nor empty.

The spam caused the sending account to be shut off, causing other
mail (routine and urgent) to be shut off with it.