Power Infrastructure Incident in VIT

Incident Report for IONOS CLOUD

Postmortem

What happened?

On July 9, 2026, the Logroño datacenter experienced a cascading power and cooling failure impacting customer workloads and services. The incident began at 21:59 CEST and required on-site manual intervention to restore service. Full power was restored at 23:05 CEST; complete service normalization concluded at 09:22 CEST on July 10.

No customer data loss has been identified.

How was this possible? (Root Cause)

The outage was caused by a single electrical fault that triggered a cascading sequence of automated safety responses.

At 21:59 CEST, a residual current device (RCD breaker D1) in the high-voltage control cabinet tripped. At this point, no visible outage occurred. However, the trip silently disabled the generator control system - the generators lost visibility of their circuit breakers and were no longer able to start automatically. The facility appeared fully operational, but its backup power capability was gone.

23 minutes later, at 22:22 CEST, the 66 kV high-voltage grid supply failed. The protection system detected overcurrent and initiated an automatic safety shutdown - a designed safety response, not a malfunction. Because the generators had been silently disabled since 21:59, they were unable to take over the load. All UPS systems switched to battery mode, protecting the workload throughout. However, as cooling infrastructure is not UPS-backed, all chillers and CRAH units lost power immediately.

Manual intervention by on-site staff restored grid power at 22:50 CEST. At 23:05 CEST, all generators were reset and the transformer breaker was closed, fully restoring power to the facility. The sequential service recovery process then began.

The exact cause of the initial D1 trip is still under investigation by our electrical engineering team and an external vendor performing forensic analysis on-site.

What we are doing to prevent recurrence

The investigation has already identified the architectural root cause: a single RCD breaker acting as a shared upstream dependency for both the generator control system and the high-voltage protection system. The following measures are being implemented:

Already completed

  • Generator validation: All diesel generators have been confirmed synchronized and operationally ready in auto-mode following the corrective actions applied on-site.
  • Infrastructure validation: On July 13, 2026, a scheduled maintenance shutdown by grid operator Iberdrola provided a real-world validation. All systems performed as designed — generators started automatically, no service interruption occurred.
  • High-voltage station vendor inspection completed — no additional defects found.

Short-term - within 4 weeks

  • Redundant AC supply for HV protection and generator control system — eliminating the identified single point of failure so that no single component can simultaneously disable both primary and backup power paths.
  • Review of RCD sizing and placement in the HV auxiliary circuit.
  • Network failover review: engineering is reviewing route reflector and load balancer failover behavior to prevent static connectivity conditions that compounded recovery time.

Structural

  • Redundant AC supply for HV protection and generator control system — eliminating the identified single point of failure so that no single component can simultaneously disable both primary and backup power paths.
  • Emergency runbook update: HV protection system failure scenario added and distributed to all on-site and on-call staff.
  • Full electrical architecture audit: a comprehensive review of the Logroño facility's electrical distribution layout to identify and remediate any further single points of failure.
  • Full post-incident review (PIR) to be completed and shared upon request.

Closing remarks

Despite the resilience measures already in place, service disruption could not be completely avoided during this event. The cascading nature of the failure - where a single circuit trip silently disabled backup power capability before the grid failure occurred - is a serious architectural finding, and we have moved immediately to eliminate that dependency.

On July 13, 2026, our infrastructure was put to a real-world test during a scheduled Iberdrola maintenance shutdown. All systems performed as designed. We are confident that the corrective actions taken are effective.

We recognize the impact this had on customers relying on our services and are committed to closing all identified action items with urgency and transparency.

Posted Jul 22, 2026 - 09:59 UTC

Resolved

We are marking this incident as resolved. The Root Cause Analysis will be published here once compiled.
Posted Jul 10, 2026 - 07:59 UTC

Monitoring

The last host and VMs have been recovered. We are setting this incident into monitoring status.

Preliminary information on the Incident:
Following a power grid outage, the site suffered a partial interruption of power, which impacted several datacenter support systems. The reason for this is under active investigation. This affected the availability of a subset of infrastructure components and hosts, subsequently impacting the VMs provisioned on them or dependent on those components.

A full Root Cause Analysis (RCA) is currently being compiled and will be shared here once completed.

Thank you for your patience.
Posted Jul 10, 2026 - 00:41 UTC

Update

All but one hosts are recovered. The remaining host suffered a hardware defect and needs to be replaced to recover the workload provisioned there. This is currently underway. We estimate that this work will be completed within the next minutes.
Posted Jul 10, 2026 - 00:28 UTC

Update

Most affected hosts have been brought back into service. We are working on restoring the remaining components.
Posted Jul 09, 2026 - 22:47 UTC

Identified

Power has been restored and site support systems are back in operational state. We have started recovering infrastructure components to fully restore service.
Posted Jul 09, 2026 - 22:29 UTC

Investigating

We are currently responding to an incident in our VIT datacenter. We will regularly update this status page with information.
Posted Jul 09, 2026 - 22:16 UTC
This incident affected: Location ES/VIT (Compute, Cubes, Storage, Network, Provisioning).