Running on OVHcloud?

OVHcloud quota or bare-metal stuck? We sort it out.

A Public Cloud project is hitting a vCPU quota that nobody raised during the move from the old datacenter, a bare-metal server at Strasbourg has not pinged since the overnight maintenance window, and the Object Storage bucket in GRA is returning 403s for a role that worked yesterday. We open the OVH API, read the task history on the server, and bring the estate back under control without waiting on a support ticket cycle.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

ovh-manager.log

still broken

What it is

OVHcloud at a glance

OVHcloud, originally OVH, is a French cloud provider founded by Octave Klaba in 1999 in Roubaix. It runs more than thirty datacenters across Europe, North America and Asia, builds its own servers, and operates its own backbone. The product line splits across Bare Metal (dedicated servers rented by the month, often at well below hyperscaler prices), Hosted Private Cloud (VMware stacks in OVH facilities), Public Cloud (an OpenStack-based service for VMs, Object Storage, Managed Kubernetes and Managed Databases), and Web Cloud (shared hosting, domains and email).

OVHcloud is the EU-sovereign alternative most regulated procurement departments reach for, and it comes with the operational style of a company that grew from colocation rather than from a developer tooling product. The API is comprehensive, the Terraform provider covers most of the estate, and the manager interface is improving but still surprises people who arrived from AWS or GCP. Trouble in production clusters around quota limits that need formal increases, bare-metal servers that silently fail hardware checks after a reboot, vRack private networking that was not attached to new instances, and scheduled maintenance windows whose timing is announced in CET and read in Pacific time.

Type
Bare metal, Public and Hosted Private Cloud, Web Cloud
First released
1999
License
Proprietary service, open-source API clients and Terraform provider
Maintained by
OVH Groupe S.A. (OVHcloud)
Typically used for
EU-sovereign hosting, bare-metal workloads, OpenStack-based Public Cloud, VMware migration targets, cold storage

The honest version

OVHcloud pros and cons

No technology is all upside. This is where OVHcloud earns its place and where it tends to cause trouble.

Where OVHcloud is strong

  • Bare metal is a first-class productDedicated servers with current AMD EPYC or Intel Xeon processors are rented by the month at prices that hyperscalers struggle to match, with real network bandwidth included.
  • EU data sovereignty is genuineDatacenters in France, Germany, Poland, the UK and other European countries, operated by a European company under European law, which matters for GDPR-heavy and public-sector customers.
  • OpenStack under the Public CloudThe Public Cloud API is OpenStack-compatible, so tools like openstackclient, Terraform and custom automation written for OpenStack keep working without a vendor-specific rewrite.
  • vRack for private networkingA single private layer-2 network can span dedicated servers, Public Cloud instances, Hosted Private Cloud and additional sites, which is unusually flexible for a cloud at this price point.
  • Cold storage is affordableOVHcloud's Cold Archive and Cold Storage products are priced well below hyperscaler equivalents, which suits backup, compliance archives and legal hold workloads.
  • Direct contact with a humanEnterprise customers can get named account managers and phone-based support in a way that the hyperscalers reserve for very large contracts.

Where OVHcloud bites

  • The manager interface has layersThe web console has grown over twenty years, and screens for different product lines look and behave differently. Engineers coming from AWS expect one coherent surface and find several.
  • Quota increases are a formal processRaising vCPU, RAM or disk quotas on Public Cloud often needs a support interaction rather than a self-service click, which is a surprise during a time-pressured scaling moment.
  • Bare-metal provisioning is slower than cloud VMsA dedicated server order can take hours rather than minutes, especially for less common SKUs, and customization at install time is more rigid than a cloud-init-driven VM.
  • Managed services are more limited than AWSManaged Databases, Managed Kubernetes and Object Storage exist and work, and they do not have the depth of RDS, EKS or S3. Teams depending on those features need to plan around the gaps.
  • Maintenance windows are in CETAnnouncements for planned maintenance arrive in Central European Time and apply to European datacenters. US teams routinely miss the timing and have a server reboot during their business hours.
  • The SBG2 fire still echoes in procurementThe 2021 Strasbourg datacenter fire destroyed customer data and prompted a long conversation about backup and resilience inside OVH. Policies improved; the memory is why most teams now treat OVHcloud backups as the start of the plan, not the end.

Sound familiar?

Common OVHcloud problems we fix

The OVHcloud problems that bring people to us, and what you have once each one is fixed.

Quota exceeded for cpu in region GRA9

Public Cloud quotas are blocking scale

A scaling event that needs more vCPU than the project is allowed, with the increase waiting on a support ticket that nobody filed in advance.

Quotas raised and documented before they bite

task hardware-check failed

A bare-metal server failed its hardware check

A dedicated server that rebooted into a diagnostic state because a RAID controller, a drive or a power supply reported a fault. The box is offline until intervention by OVH staff.

Hardware investigated and recovered

vRack: private network not attached

New instances are missing the private network

A vRack that was set up for the first wave of resources and not attached to instances created later, so internal services cannot reach each other even though the external IPs work.

vRack attachments documented and enforced

403 Forbidden on s3:PutObject

Object Storage permissions drifted

A user whose role or S3 credentials no longer match the bucket policy after a rotation, or a region that was moved from the old Public Cloud to the new Object Storage product without updating the clients.

Credentials and policies aligned

cloud-init: user-data script failed

Cloud images do not accept user data

An OVH image that ships without cloud-init or with a stripped version, so Terraform or packer workflows that assume cloud-init silently fall through without running the bootstrap.

Images that execute user-data cleanly

scheduled maintenance SBG1

A maintenance window took a server offline at a bad time

A notice announced in CET that arrived in a mailbox nobody reads, or a window that was accepted months ago and had been forgotten. The reboot interrupts a job that nobody protected.

Maintenance windows calendared and routed

Managed Kubernetes: context deadline exceeded

OVH Managed Kubernetes is slow to respond

A control plane that timed out during a node pool operation, or a Terraform plan against the cluster that hit the API window. Operations partially succeed and state drifts.

Terraform retries and timeouts tuned

no runbook, inherited estate

We took over an OVHcloud account with no handover

Dedicated servers under personal emails, Public Cloud projects on an ex-employee's login, and billing on a card nobody has. The account works until it does not.

Account recovered, keys rotated

What we do

OVHcloud help, from rescue to steady running

Reach out for one locked server, or let us look after the whole OVH estate.

Rescue
Bare-metal servers that failed hardware checks, Public Cloud quotas that broke a deploy, or Object Storage buckets refusing writes. We stabilize first and explain the cause second.
Build-out
A new OVHcloud estate laid out in Terraform with vRack, Object Storage, Managed Kubernetes and backup policies, so nothing is clicked into existence without a record.
Migrations
Moving workloads from AWS, GCP or Azure onto OVHcloud for sovereignty or cost reasons, with database replicas streaming across and DNS cutovers planned.
Bare-metal plus Public Cloud hybrid
Databases on dedicated servers wired through vRack to elastic Public Cloud instances, keeping bare-metal economics without losing scale-out capability.
Account recovery
Taking over OVH accounts still under former employees, rotating API keys, moving NIC ownership, and auditing every active resource.
Managed Tech
Patching, backup verification, maintenance window tracking, and on-call cover for the night a Strasbourg rack decides to reboot earlier than announced.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running OVHcloud alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

OVHcloud questions

Something else? Call+1 737 727 7327

Is OVHcloud reliable enough for production?

For a very large number of European and global workloads, yes. OVH runs datacenters that have been up for years, operates its own backbone, and has credible network engineering behind the scenes. The 2021 SBG2 fire is the honest counterpoint, and the response was to tighten fire suppression, cross-site replication and backup policies. We treat OVHcloud as a serious production platform and still insist on backups into a second provider or a second region for anything that matters.

Why can we not scale our OVHcloud Public Cloud project?

Almost always a quota. vCPU, RAM, disk, floating IPs and Managed Kubernetes clusters all have per-project quotas that start low and rise on request. The manager will show the limit, and the increase is a support interaction rather than a self-service toggle, so the usual fix is asking for a raise before the scaling moment arrives rather than during it.

How do we handle maintenance windows on OVHcloud?

Announcements arrive in CET and in advance, often weeks before the window. We route those notifications into a shared calendar and a Slack channel rather than a personal mailbox, confirm the affected servers in the manager, and either accept the window or request a reschedule where it conflicts with a change freeze. Treating them as planned incidents rather than surprises is the difference.

Can you move us from AWS or Azure to OVHcloud?

Yes. The usual driver is cost, sovereignty, or both. We build the OVHcloud side in Terraform, stream the databases across in advance, cut DNS over during a quiet window, and keep the old provider in standby mode for a week or two before decommissioning. Backup into a second region or a second provider is part of the plan from day one, not an afterthought.

Can you take over an OVHcloud account from a previous team?

Yes. We work through the NIC transfer, rotate every API key and server credential, move billing onto a company card, and inventory what is running across dedicated servers, Public Cloud projects and vRack networks so the handover is clean and nothing goes dark silently.

How much does OVHcloud help cost?

Quotes are free. Tell us the regions, which products you are running, and what is going wrong, and we return a fixed price before any changes touch the account.

Free estimate

OVHcloud giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.