Notice a missing heartbeat

Save the following as heartbeat.yaml and validate it with the watch preview. This example needs no external provider. It prints an alert if ten minutes pass without fresh input and recovers when another observation arrives.

apiVersion: ding.ing/v1alpha1
kind: Destination
metadata: {id: console}
spec: {type: console}
---
apiVersion: ding.ing/v1alpha1
kind: Watch
metadata: {id: heartbeat, name: Job heartbeat}
spec:
  source: {type: push, fields: {alive: alive}}
  condition: {missingFor: 10m}
  policy: {trigger: transition, consecutive: 1, recoverAfter: 1}
  destinations: [{ref: console, events: [firing, recovered]}]

Download the manifest.

./ding validate heartbeat.yaml --json
./ding apply heartbeat.yaml --state-dir ./ding-state --dry-run --json
./ding apply heartbeat.yaml --state-dir ./ding-state

Send {"alive":true} to /v1/ingest/heartbeat using the ingest token, as in the push guide. The ungrouped deadline starts when applied. Unknown input does not postpone it; HTTP 304 can count as freshness for an HTTP missing-data watch.

The deadline persists through restart and fires once until recovery. Pausing removes deadlines while preserving incident state; resuming starts fresh deadlines. For a grouped watch, a deadline begins only after each entity is first observed. Ding cannot detect an entity it has never seen.