Summary
- OVHcloud’s Object Storage S3 service in the GRA region was disrupted from 02:00 UTC on 3 October.
- Customers experienced HTTP 500 and 503 errors, with deletion requests particularly affected.
- OVHcloud traced the incident to the loss and instability of two servers and restored service by 04:27 UTC.
OVHcloud has restored Object Storage S3 services in its Gravelines region after a hardware fault involving two servers caused errors for customer requests early on Saturday.
The incident began at 02:00 UTC on 3 October and affected both standard and high performance Object Storage S3 services in the GRA region.
OVHcloud said customers could experience HTTP 500 and 503 errors when handling objects, including problems with upload and download operations. Deletion requests were particularly affected.
The company initially identified the problem as an active service incident and mobilised teams to restore the platform. Service was fully restored by 04:27 UTC, with monitoring continuing afterwards.
In its later incident summary, OVHcloud attributed the disruption to a hardware issue. The loss and instability of two servers interfered with the handling of deletion events, according to the provider.
The incident remained regional rather than becoming a general OVHcloud outage. That is important when assessing object storage resilience because distributed storage platforms are designed to tolerate individual component failures, but the behaviour of the services coordinating metadata and object operations can still expose customers to errors when hardware becomes unavailable or unstable.
Object storage also presents a different failure pattern from conventional virtual machines. Customers may still be able to reach an endpoint while only certain operations return errors. A platform can therefore remain partially functional even as applications encounter failures when attempting to write, delete, or retrieve specific data.
For customers, HTTP 500 and 503 responses usually require retry logic at application level. The effect of an incident therefore depends partly on how software using the storage platform is designed. Applications with robust retry and queueing mechanisms may ride through a short interruption, while tightly coupled workflows can fail when a storage operation does not complete.
OVHcloud’s Gravelines campus is one of the company’s major French infrastructure locations, making the regional service stack an important part of its European cloud footprint. Saturday’s issue was resolved in a little over two hours from the recorded start time and OVHcloud did not report data loss in its incident notice.
The provider has not disclosed why the failure of the two servers affected deletion event handling or whether additional changes are being made following the incident. The published root cause statement is therefore narrower than a full technical review.
The event nevertheless demonstrates why cloud availability cannot be reduced to whether a data centre retains electricity or network connectivity. A facility can remain operational while a storage layer inside it becomes partially unavailable because of hardware or software dependencies lower in the service stack.
With service restored, the operational follow up will be whether OVHcloud publishes further remediation detail or identifies changes intended to prevent a repeat of the same failure mode.
