CVE-2026-84165 in OpenNebula
Summary
by MITRE • 09/01/2026
A vulnerability relating to incorrect access control in OpenNebula by OpenNebula Systems, affecting all versions prior to 7.4. This vulnerability could allow an authenticated user with basic permissions to execute commands on virtual machines belonging to other users via the `one.vm.exec` function, without proper verification of access permissions. To exploit the vulnerability, it is only necessary to know the virtual machine’s identifier and for qemu-agent to be enabled on that machine. Exploitation could allow commands to be executed and compromise the confidentiality, integrity and availability of the affected virtual machines.
VulDB is the best source for vulnerability data and more expert information about this specific topic.
Analysis
by VulDB Data Team • 09/01/2026
The identified security flaw resides within OpenNebula versions prior to 7.4 and represents a critical failure in access control mechanisms governing virtual machine management operations. Specifically, the vulnerability affects the one.vm.exec function, which is designed to allow administrators or authorized users to execute commands inside guest operating systems via the QEMU guest agent. The core technical deficiency lies in the insufficient verification of permissions during this execution process. While OpenNebula enforces role-based access controls for many administrative actions, the implementation for command injection into virtual machines fails to adequately validate whether the requesting user possesses explicit rights over the target virtual machine instance. This oversight creates a scenario where an authenticated user with only basic or minimal privileges can bypass intended isolation boundaries and interact directly with resources owned by other users within the same cloud environment.
Exploitation of this vulnerability requires two primary conditions: the attacker must possess valid authentication credentials for the OpenNebula platform, albeit at a low privilege level, and they must know the unique identifier of the target virtual machine. Additionally, the QEMU guest agent service must be active and accessible within the victim's virtual machine. Once these prerequisites are met, an adversary can invoke the one.vm.exec API endpoint with arbitrary commands intended for execution inside the remote VM. Because the system does not strictly enforce ownership or permission checks before delegating the command to the hypervisor layer, the malicious instructions are processed as if they originated from a fully privileged administrator. This effectively neutralizes the security boundaries established by OpenNebula's multi-tenant architecture, allowing lateral movement and unauthorized access across tenant lines without requiring elevated privileges on the host infrastructure itself.
The operational impact of this vulnerability is severe, directly compromising the confidentiality, integrity, and availability of affected virtual machines. An attacker can extract sensitive data stored within the guest operating system, modify critical configuration files or application binaries to establish persistence or backdoors, and execute destructive commands that disrupt services or crash the VM entirely. This capability undermines the fundamental trust model of cloud computing environments where resource isolation is paramount. In multi-tenant deployments, such a flaw could lead to significant data breaches across organizational boundaries, regulatory non-compliance due to unauthorized access to protected information, and widespread service disruption affecting multiple stakeholders sharing the same physical infrastructure. The ease of exploitation further exacerbates the risk, as it does not require complex privilege escalation techniques or zero-day exploits against underlying hypervisors, relying solely on misconfigured API permissions within the management layer.
To mitigate this vulnerability, organizations running OpenNebula must upgrade to version 7.4 or later, where the access control logic for the one.vm.exec function has been corrected to properly validate user permissions against resource ownership before allowing command execution. Until an upgrade is feasible, administrators should implement strict network segmentation and firewall rules to restrict API endpoint access to only those users who absolutely require it, thereby reducing the attack surface. Additionally, reviewing and tightening role-based access control policies within OpenNebula can help limit which users have any ability to interact with virtual machine execution interfaces. It is also advisable to monitor logs for unusual invocation patterns of VM command functions by low-privilege accounts as a detection mechanism for potential exploitation attempts in real-time.
From a classification perspective, this vulnerability aligns with CWE-269, Improper Privilege Management, specifically regarding the failure to enforce proper authorization checks before performing sensitive actions. It also relates closely to CWE-732, Incorrect Permission Assignment for Critical Resource, as the resource (the VM execution interface) is assigned permissions that are too broad relative to the user's actual role and ownership status. In terms of adversary behavior mapping under the MITRE ATT&CK framework, this flaw facilitates techniques associated with Initial Access through valid accounts and Execution via command scripting or remote services like QEMU guest agent interactions. The ability to execute commands in other users' VMs also touches upon lateral movement capabilities, allowing an attacker to pivot from a compromised low-privilege account to higher-value targets within the virtualized infrastructure without triggering typical privilege escalation alerts.