Spotting data gaps in history

Recognize missing inserts, feed outages, and reboot gaps in dashboard charts before they skew freeze audits.

Home desk with laptop, external monitor, and keyboard in a living space
Household members check the same dashboard from wherever they already work. Photo: David Wellbeloved / Flickr. CC BY 2.0

Saved history is only as complete as the polling pipeline. Power loss at the Arduino, relay downtime, auth misconfiguration, or Supabase insert failures all leave gaps. Learning to distinguish **missing data** from **flat temperature** prevents false confidence during cold snaps and billing disputes over CSV completeness.

Chart with a visible gap versus a flat line segment labeled differently
True gaps show missing timestamps; flat segments may be valid stable readings: learn the visual difference.

Common gap causes

  1. Feed unreachable: relay down or IP change; start with stale reading debug.
  2. MCU reboot: short blank period after watchdog reset.
  3. Insert skipped: upstream fetch failed per history insert rules.
  4. Account/session: signed-out loads do not save; see session lifecycle.

Visual patterns

  • Vertical whitespace in charts when time axis jumps forward.
  • Repeated identical values can mean stuck cache: check feed caching.
  • Single-zone flat while others move suggests mapping or sensor fault.

Workflow

Cross-check live JSON with curl, then browse history dashboard. Export via CSV subscription and filter timestamp column for discontinuities: techniques in spreadsheet charting.