Summary
On September 9, 2026, at approximately 23:44, we experienced a loss of upstream connectivity at our Copenhagen location. This resulted in network connectivity issues affecting parts of our Public Cloud platform as well as several supporting services, including our API, dashboard, DHCP infrastructure, and one of our authoritative nameservers.

The incident was caused by a loss of the physical link between our upstream provider's CPE and our edge network. As a result, our BGP sessions to the upstream provider were dropped and external connectivity through this link was lost.

Customers using transit services with redundancy provided by us were not affected. The impact was limited to services and customers relying solely on the affected upstream path.

Two technicians were dispatched to the datacenter during the night. The physical link was restored at approximately 04:44. Additional troubleshooting was subsequently required because the BGP sessions did not immediately re-establish after the physical connection had been restored.

BGP connectivity was successfully restored at approximately 05:15, after which routes and services gradually converged. All affected services were fully operational again by 05:30.

Root Cause

The primary cause of the incident was the loss of the physical connection between our upstream provider's CPE and our edge routers.

This physical connectivity loss caused the BGP sessions between our network and the upstream provider to drop, resulting in the withdrawal of routes normally announced through this connection.

Our initial investigation identified the missing BGP sessions as the immediate cause of the loss of connectivity. Further investigation showed that the underlying problem was at the physical/link layer.

Once technicians restored the physical link, BGP sessions did not immediately recover as expected. Additional changes and coordination with our upstream provider were required before the sessions could successfully be re-established.

Services are fully restored now

New BGP sessions has been established to our upstream, we are monitoring the network slowly restoring

The link has now been fixed, we are still seeing issues with BGP sessions not establishing

Our two technicians has now arrived on site

We have two technicians on the way to the site to investigate the physical link and to perform some requested changes made by our upstream.

Update to impact: Transit customers with redundancy provided by us are not affected. The impact is limited to customers without redundancy.

Its still being investigated, currently it seems like the link between the CPE from our upstream and our edge routers are down.

This might require a physical technician to perform work, in that case we will be sending one out in the morning if needed.

Update: We lost full upstream link

We have established the root cause is dropped BGP sessions, we are in talks with our upstream provider to resolve the issue.

We are investigating upstream connection loss

API and Public Cloud went down.