Summary
- All servers in rack RBX1002B04C became unavailable at 22:20 UTC on 28 September.
- OVHcloud identified a cooling-system issue as the root cause.
- Servers began returning by 22:51 UTC and the incident ended at 22:54 UTC.
OVHcloud temporarily lost every dedicated server in one rack at its RBX10 facility in Roubaix after a cooling-system fault, according to the operator’s incident record.
The event began at 22:20 UTC on 28 September and affected rack RBX1002B04C. OVHcloud said all servers in the rack were temporarily unavailable and customers could not access machines located there.
The operator identified the root cause as a cooling-system issue. By 22:51 UTC some servers were progressively returning to operation, and the incident was formally closed at 22:54 UTC.
The customer-impact window therefore lasted 34 minutes. OVHcloud did not provide further detail on the failed cooling component, temperatures reached inside the rack, or whether servers shut down automatically or were taken offline by operators.
Rack-level cooling becomes an availability issue
The narrow scope should not be overstated: OVHcloud did not report a building-wide cooling failure, and the incident affected a specified rack rather than the entire RBX10 facility.
It nevertheless demonstrates the direct link between thermal infrastructure and service availability. Servers can have redundant network paths and power supplies but still become unavailable if the cooling system protecting their operating environment fails.
Higher rack densities are making that relationship more pronounced. As more power is concentrated into a smaller footprint, the time available to react to reduced airflow, chilled-water problems, or other thermal faults can shrink.
OVHcloud conducts regular cooling maintenance across its estate, and its status systems list numerous planned interventions on individual racks and facilities. Those maintenance records should not be conflated with the unplanned RBX10 event, but they show how frequently mechanical infrastructure requires operational attention.
The incident also follows a separate OVHcloud service disruption in Paris earlier this month. DataCentral reported on that network-infrastructure malfunction after service was restored. The Roubaix event had a different cause and a much smaller footprint.
Rack-level incidents can expose dependencies that defeat redundancy locally. Cooling distribution, sensors, valves, fans, controls, and maintenance procedures all sit between central plant resilience and the actual thermal conditions experienced by servers.
OVHcloud restored the affected equipment quickly, and no wider impact was reported. The company has not published a fuller post-incident analysis, so the available evidence supports only the identified cooling-system cause and the published restoration timeline.
