Status

9 January 2026, 07:58–08:45 UTC · Live status

Partial position data for European airspaceResolved

Position updates for flights over parts of Europe lagged by up to 11 minutes (normal is 4–9 seconds) after an ADS-B receiver-network partner lost connectivity during a datacenter migration. REST status fields stayed correct but stale for position, altitude and speed; everything else — schedules, gates, delays — was unaffected. Customers polling position saw old-but-plausible data rather than errors, which is the most annoying kind of incident: nothing alerted, because the endpoint was technically fine.

Detection and fix

• A customer's freshness monitor caught it at 08:10 UTC (our own alerting watched error rates, not data age — see below).
• Traffic failed over to a secondary receiver feed at 08:31; freshness normalized by 08:45.
• Historical position data for the window was backfilled from the secondary feed; replay is available to Business plans.

Changes

• We now alert on ingestion lag per region, not just availability. This incident class can't hide again.
• Added position.stale boolean to flight objects so consumers can distinguish "fresh" from "best available."

20 November 2025, 16:05–19:40 UTC · Schedules

Schedules up to six hours staleResolved

A carrier change file with a malformed date (20250230 — a valiant attempt at a February 30th) crashed the schedules importer mid-run. The importer was supposed to skip bad rows; it instead stopped entirely, and 140 carriers' updates for the day queued behind it. Served data stayed valid but aged: several airlines' schedules reflected Thursday instead of Thursday evening.

Fix and follow-up

• Bad-row quarantine and importer restart completed by 19:40 UTC; all queues drained by 21:00.
• We called the carrier, who fixed the file at the source. Everybody was very nice about it.
• Importers now run per-carrier, so one airline's imaginary calendar can't hold up 139 others.