A network connectivity issue with an upstream provider meant that all requests to a small number of portals were failing. Automated monitoring initiated fail-over for the affected component, and all portals are online as normal.
Partial outage first detected 06:47. Resolved 06:51, and all portals online within 3 minutes of resolution.
We are investigating alternative approaches to allow more immediate reaction to network failures.
Apologies for the disruption.