PulseHeberg
← Back to the status page

Incident · Minor impact

Incident ponctuel sur un nœud de calcul à Marseille

Affected components : Virtual servers (Partial outage), 🇫🇷 Paris-Magny (PAR) (Partial outage)

  1. Resolved

    The very brief incident affecting a compute node in Marseille is now resolved.

    Our system automatically failed over the affected VPS to another healthy node, which may have caused an unplanned restart of the guest systems. The affected node was then removed from production in accordance with our procedure.

    All affected VPS in the Economy Cloud and Standard Cloud ranges are operational again. In the vast majority of cases, no action is required on the part of customers. However, should an issue persist, we ask affected customers to restart their VPS.

    We apologize for the inconvenience.

  2. Investigating

    We are currently investigating an incident affecting a compute node in Marseille. Some VPS on the Economy Cloud and Standard Cloud ranges may be unavailable or experiencing access difficulties.

    Our teams are working to restore the affected services as quickly as possible. A further update will be published once we have more information.

Started
Resolved

Post-incident review

In the early hours of 20 September 2026, an isolated failure affected a compute node in our Marseille infrastructure. Some VPS instances in the Economy Cloud and Standard Cloud ranges hosted on this node became unavailable at around 01:25 Paris time.

Our system detected that the node was unavailable and automatically triggered failover of the affected virtual machines to other healthy nodes. The VPS instances were then restarted progressively, with all services restored at around 02:20.

This failover explains the unplanned restarts observed by customers, as well as the absence of a normal shutdown sequence in the guest operating system logs.

The affected node was immediately removed from production in accordance with our procedure and will remain isolated until all technical checks have been completed. The affected VPS instances are now running on healthy infrastructure.

Our teams verified that the virtual machines had restarted correctly, checked their connectivity, and confirmed the health of the infrastructure hosting them. All systems are operating normally again, and no remaining issues have been identified.

This event was limited to a single compute node. It was not planned maintenance, and no action is required from customers.

We apologise for this interruption.

Incident ponctuel sur un nœud de calcul à Marseille | PulseHeberg Status