Issue with InfluxDB becoming unresponsive after daylight savings time

After our clocks turned back on 11/6, InfluxDB was not properly tracking all alerts in the queue. After a new alert was created, it was undiscoverable by our incident application via the API, lost the heartbeat. It is believe this was due to the fall back.

The issue names no files, tests, or entry points. Start by reproducing alert creation and API lookup across the fall-back daylight-saving transition, then trace queue tracking and heartbeat behavior. Done means newly created alerts remain discoverable through the API and retain their heartbeat after the clock change.

Written by the indexing model from the issue text.