The nexus_sources_metadata table provides one row per configured source, combining declarative ingestion config (how the source is loaded) with warehouse freshness (how recently it was loaded). It's the canonical place to answer "when did each source last get data, and how does it get here?".

Overview

Each row joins two things the warehouse alone can't give you together:

• Config side — from the optional ingestion: block on each source in var('nexus').sources: the pipeline (connector), the ingested_at_field (the raw column mapped to _ingested_at), and the source's enabled/events/dimensions/measurements/entities/attribution flags.
• Warehouse side — MAX(_ingested_at), MIN/MAX(occurred_at) and event count per source, grouped from nexus_events.

A FULL OUTER JOIN surfaces both configured-but-empty sources (is_in_data = false) and sources present in the event log that aren't configured here (is_configured = false).

Freshness semantics

last_ingested_at is MAX(_ingested_at) — the real warehouse load time for the source's ingested_at_field (Airbyte's _airbyte_extracted_at, an ETL's etl_load_dt, etc.). Because MAX ignores NULLs, sources that populate their load column only sparsely still report their true latest load.

When a source has no warehouse load column at all (e.g. a GA4 export that only carries event timestamps), its _ingested_at is NULL and last_ingested_at is NULL — a genuine "unknown", never a fabricated build-time value. Consumers should treat NULL as unknown and may fall back to last_event_at (the latest occurred_at) as a data-recency proxy.

Schema

Field	Type	Description
source	String	Canonical source identity (the source value in nexus_events)
config_key	String	The key under var('nexus').sources (may differ from source)
pipeline	String	Ingestion connector/pipeline, e.g. airbyte, cinch_etl, custom
ingested_at_field	String	Raw column mapped to _ingested_at (e.g. _airbyte_extracted_at); NULL if none
enabled	Boolean	Source enabled in config
has_events	Boolean	Source contributes events
has_dimensions	Boolean	Source contributes event dimensions
has_measurements	Boolean	Source contributes event measurements
has_entities	Boolean	Source contributes entities
has_attribution	Boolean	Source participates in attribution
last_ingested_at	Timestamp	MAX(_ingested_at) — real load time, or NULL if unknown
first_event_at	Timestamp	MIN(occurred_at)
last_event_at	Timestamp	MAX(occurred_at) — data-recency proxy when load time is unknown
event_count	Integer	Number of events from this source in nexus_events
is_in_data	Boolean	Source has rows in nexus_events
is_configured	Boolean	Source is declared in var('nexus').sources

Configuring ingestion metadata

pipeline, ingested_at_field, and an optional source_label come from an ingestion: block per source. All keys are optional and backward-compatible — omit them and the columns are NULL.

vars:
  nexus:
    sources:
      google_ads:
        events: true
        ingestion:
          pipeline: airbyte
          ingested_at_field: _airbyte_extracted_at
          # Optional: the nexus_events.source literal to join on, when a
          # source's emitted `source` value differs from its config key.
          # Defaults to the config key.
          source_label: google_ads

Query Examples

Source freshness, stalest first

select
    source,
    pipeline,
    ingested_at_field,
    last_ingested_at,
    event_count
from {{ ref('nexus_sources_metadata') }}
order by last_ingested_at asc nulls first

Sources with an unknown ingestion time

select source, pipeline, last_event_at
from {{ ref('nexus_sources_metadata') }}
where last_ingested_at is null
  and is_in_data