Summary
- UpCloud reported a File Storage backend issue at FI-HEL2 beginning at 14:01 UTC on 28 September.
- Some NFS services experienced unwanted failovers, prompting additional controlled failovers to isolate the problem.
- A fix entered monitoring at 15:34 UTC and the incident was declared resolved at 16:41 UTC.
UpCloud has restored File Storage services at its FI-HEL2 data centre in Helsinki after a backend fault caused unwanted NFS failovers and led the operator to carry out additional failovers to isolate the problem.
The incident began at 14:01 UTC on 28 September. UpCloud said some File Storage services were affected and that its systems had registered unwanted failovers on a number of NFS services.
The operator responded by performing further failovers as a safety measure, with the aim of isolating the remaining issue and returning the backend to its normal state. That intervention meant additional NFS services could experience failover activity while remediation was under way.
A fix had been implemented by 15:34 UTC, when the incident moved into monitoring. UpCloud declared the event resolved at 16:41 UTC, giving the published incident a duration of two hours and 40 minutes.
Failover does not remove operational risk
The status record does not identify the underlying hardware or software failure, nor does it specify how many customers or storage systems were affected. It therefore does not support conclusions about the precise fault mechanism.
What it does show is the operational trade-off around redundant storage. Failover is designed to preserve service when a component or backend becomes unavailable, but an unintended failover can itself become an operational event if systems move state unexpectedly or repeatedly.
UpCloud’s decision to trigger further controlled failovers illustrates the distinction between redundancy existing in an architecture and operators being able to use that redundancy safely during a fault.
NFS-based file services can support applications that expect persistent shared storage. Instability at the storage layer can therefore affect workloads even when compute instances and the wider data centre remain operational.
UpCloud classified the incident as affecting FI-HEL2 File Storages rather than the entire Helsinki facility. Other published components at the site were shown as operational, and the company did not report a complete data centre outage.
The limited scope is important. A backend storage failure is not equivalent to a facility-level power or cooling incident, even though both ultimately surface to customers as an availability problem.
The event nevertheless provides a concise example of why operational resilience depends on more than spare equipment. Detection, automation, failover logic, runbooks, and the ability to isolate a malfunctioning backend determine whether redundancy contains an incident or spreads disruption between components.
UpCloud has not published a post-incident report beyond the status updates reviewed for this draft. Any deeper account of the fault should therefore wait for additional technical information rather than inferring a root cause from the failover behaviour alone.
