How to tell if my gateway is still alive — reading the import health line

Every import that has an ongoing fetcher behind it — a scheduled API connection or a webhook source — shows a health line directly under its status badge on Analytics → Data Import. Read that line first: it tells you whether a missing-data problem is on the sending side or in the platform's own fetching.

Where to find it: Analytics → Data Import → the row for your import. The Status column stacks three things: the status badge, the last error message (in red, if there is one), and the health line.

What the health line says

• "Auto-fetch OK · 2 hours ago" — the fetcher last succeeded at that time. Judge this against the cadence you expect (see below), not against a fixed number of hours.
• "Auto-fetch retrying (3×)" — recent attempts failed but the fetcher is still running and backing off. Hover the line to see the last error.
• "Auto-fetch failing (7×)" — repeated failures. A "Retry now" action sits next to it, which clears the failure counter and fetches immediately.
• "Auto-fetch paused" — the fetcher is not scheduled at present.
• "Auto-fetch disabled — API key rejected" — the source kept answering 401/403, so the connection was switched off to stop hammering it with bad credentials. Fix the credential at the vendor, then use "Re-enable".
• No health line at all — either the import has no ongoing fetcher (a one-off file upload), or the fetcher exists but has never run yet.

Judge staleness against the cadence you expect, not a fixed age.

A gateway configured to push hourly is already broken after two or three missed pushes. A monthly utility report is perfectly healthy at 25 days old. There is no single "too old" threshold that fits both — compare "Auto-fetch OK · X ago" with the interval you configured on the import (or the schedule the sender was set up with).

The status dot in the Analytics header is a coarse backstop. It turns amber and reads "Data delayed from N source(s)" only when a whole source type has produced nothing for more than seven days, and red when the analytics service itself is unreachable. It is a last-resort alarm for sources that have gone completely silent — a green dot does not mean an hourly feed is on time.

Diagnostic patterns

• Fetches succeed until a specific time, then stop: SIM data plan exhausted, server certificate distrusted by the gateway, secret rotated, or the gateway lost power. Check the gateway physically.
• Success alternating with failures: intermittent signal (move/extend the LTE antenna), or the gateway clock is drifting in and out of the 5-minute tolerance window (enable NTP).
• Only failures, never a success: signing logic or URL is wrong from day one. Run a curl test from a laptop with the same secret to isolate.
• Healthy line but readings still missing from the dashboard: the discovery step hasn't been completed — see "My webhook returns 200 but no data appears in dashboards".

What the platform does NOT show you

• A per-delivery log. You get the last error and the consecutive-failure count, not a history of every request.
• The body of rejected requests (they are not stored).
• The specific signature mismatch reason (avoiding leaking attack-useful detail), or the remote IP that delivered a request.

If you need that level of detail, contact support with the import name and the time range of the failures — see "What to send support when an import stops working".