How to read this
Every state is what the vendor says about itself, read from the API-facing components of its public status page. Apideck does not measure it, and vendors under-report: a connector shows as degraded whenever the vendor has an open incident on a tracked component, even while the component itself still says operational. History covers the last 90 days, or less where the vendor’s feed caps at its 50 most recent incidents; the hatched part of a timeline is the gap.
During a vendor incident
Unify returns the vendor’s failure as a ConnectorExecutionError, with the vendor’s own status code and message in downstream_errors:
{
"status_code": 503,
"type_name": "ConnectorExecutionError",
"message": "QuickBooks Connector Returned Error",
"downstream_errors": [{ "code": 503, "message": "Service Unavailable" }]
}If those started when the incident opened, the incident is the cause. Treat 5xx and timeouts as transient: retry with exponential backoff for as long as the vendor reports the incident open, and contact us if failures continue after the vendor resolves it.
