Xen Orchestra VSA-2026-062: patch and tidy the admin network
Xen Orchestra Vates XCP-ng CVE sécurité VSA-2026-062 administration multi-hyperviseur

Xen Orchestra VSA-2026-062: patch and tidy the admin network

Gaël Simonneaux · 13/11/2026

On 10 November 2026 Vates published VSA-2026-062, an actively exploited authentication bypass in Xen Orchestra. The versions to deploy, the indicators the vendor shared, and the admin-interface hygiene to come back to.

On 10 November 2026, Vates published VSA-2026-062: an authentication bypass in Xen Orchestra, actively exploited, that lets an attacker with network access to the admin interface retrieve an administrator token without any credential. Teams that expose Xen Orchestra or its Xen Orchestra Appliance (XOA) on a shared network should deploy the patch right away.

What the flaw does

The attacker crafts a cookie containing an empty authentication token (authenticationToken=j:{}) and sends it to Xen Orchestra, which returns an administrator token in the clear. From there, they operate like any Xen Orchestra administrator: they see every pool, start, stop, clone or delete machines, read backups and gain control of the XCP-ng hosts. Vates writes, in its advisory, that compromising Xen Orchestra "can lead to arbitrary code execution on managed pools and a full XCP-ng node takeover as root".

The versions to deploy

Deploy a patched version immediately:

Rolling back to an older version is never a solution: the flaw sits upstream.

Check and clean up behind the patch

Vates mentions several indicators of compromise in its advisory:

Rotation of secrets and keys stored on the affected machines, review of recent permission changes and new Xen Orchestra users, log review and backup verification within the suspected window: the usual steps after a compromise, without haste.

And beyond: what the admin interface never had to see on the internet

This Vates advisory is a reminder of a discipline that teams know, but that a busy week will nibble at: the admin interface of a hypervisor should never be reachable from the internet. It applies to Xen Orchestra and its XOA, to the Proxmox VE portal, to vCenter, to Prism Central, to Hyper-V Admin Center. A platform's safety net is its management network: a dedicated, isolated subnet, reachable only through a bastion or a VPN, with a source IP allow-list where possible.

NaoShift follows that discipline when it reads your XCP-ng pools: the naoshift agent installs on your side, talks to Xen Orchestra over your management network, and opens the dialogue with naoshift on outbound HTTPS. No inbound port to open. No packet arrives from the outside on Xen Orchestra. The same principle applies to every hypervisor NaoShift manages: read-only by default, write enabled per target with a dedicated account, traced before execution.

What NaoShift sees, and what it does not

The inventory of XCP-ng missing patches lives inside Xen Orchestra, which exposes it per pool (pool.listMissingPatches). NaoShift reads it with the collection account, names the CVEs and the XSA bulletins, and raises risks per host: at a glance, across a multi-pool estate, you see which nodes are waiting for a fix and what it corrects.

NaoShift does not read Xen Orchestra's own version, nor its cookies, nor its sessions; the responsibility for patching Xen Orchestra itself still sits with the team that runs it. Following the official list of Xen Orchestra patches is still, for now, a manual step — or one you automate through your package management. An advisory like VSA-2026-062 is a good moment to check that cadence.

Frequently asked questions

Is XCP-ng affected, or only Xen Orchestra?

The defect sits in Xen Orchestra, not XCP-ng. But compromising Xen Orchestra gives access to the XCP-ng hosts it manages: in practice, an exposed XCP-ng cluster should be treated as vulnerable until Xen Orchestra is patched.

What if the admin interface was open to the internet during the exposure window?

Patch first, then handle the compromise as a security incident: the indicators Vates shared, a review of machines created or changed outside your routine, backup verification, secret rotation, admin log review. Then close the admin network so the issue does not come back through the same door.

Does NaoShift detect exposed appliances and pools?

NaoShift inventories XCP-ng pools, host versions and missing patches. It does not probe your network from the outside and so cannot tell you whether your Xen Orchestra interface is reachable from the internet. Your network team will always know that better than any SaaS platform.

Where to go next

NaoShift manages VMware vSphere, Proxmox VE, XCP-ng, Microsoft Hyper-V and KubeVirt clusters (OpenShift Virtualization, Superphenix, HPE VM Essentials) from a single console: consolidated inventory, CVE and missing patches per host, centralized rights, exportable audit log. Start a free 30-day trial or get in touch.