CVE-2026-16812: Arista VeloCloud Orchestrator On-Prem OS Command Injection Vulnerability

CVE-2026-16812 is an OS command injection vulnerability in Arista VeloCloud Orchestrator On-Prem. A remote attacker who can reach the affected interface may abuse it to run commands in a privileged context on the orchestrator host, which can put the confidentiality, integrity, and availability of the orchestrator and the data it manages at risk.

SD-WAN orchestrators sit at the center of network control and visibility. Compromise of that plane can affect configuration, telemetry, and connectivity decisions across sites. Confirm exact scope, fixed builds, and exposure conditions against the vendor advisory before acting.

How it works

This issue is classified as CWE-78 (OS command injection). In products of this class, user-controlled input is passed into a shell or system command without sufficient validation or separation of data from code. An attacker who can supply that input—often via an authenticated or network-reachable management path—may cause the application to execute attacker-chosen operating-system commands with the privileges of the service account.

Public detail for this CVE does not describe the exact request path, parameter, or authentication requirement. In general, successful abuse of command injection on an orchestrator can yield host-level access, lateral movement into management networks, and tampering with or disclosure of configurations and operational data. Do not assume unauthenticated remote code execution or any specific exploit chain without confirmation from the vendor advisory and your own testing in a lab.

Am I affected? How to find it in your systems

Arista VeloCloud Orchestrator On-Prem is typically deployed as a central management component for VeloCloud SD-WAN environments—often in a data center, private cloud, or dedicated management VLAN, and sometimes with administrative access from the internet or partner networks. Inventory every on-premises VCO instance, including lab, DR, and forgotten appliances.

• Enumerate hosts by asset management, CMDB, network discovery, and certificate or banner checks for VeloCloud/Orchestrator management services.
• Record software version, build, and deployment mode (on-prem vs cloud) and compare them only to the fixed and affected ranges published in the vendor advisory—do not rely on third-party version lists alone.
• Note whether the management UI or APIs are reachable from untrusted networks, and which accounts or integrations can call those interfaces.
• Review authentication logs, admin session history, and process or shell audit logs on the VCO host for unexpected commands, new local users, unusual outbound connections, or configuration changes outside change windows. Specific exploitation indicators are not provided in the public summary; treat anomalies on the orchestrator as high priority until ruled out.

How to remediate

Patch first. Apply the vendor-supplied update or mitigation package for Arista VeloCloud Orchestrator On-Prem exactly as described in the official advisory. Validate the installed version after upgrade and retain evidence for change control and any regulatory obligations (including CISA BOD 26-04 prioritization where it applies to your environment).

• If the advisory offers configuration hardening or a temporary workaround in addition to the patch, implement those only as documented by the vendor.
• After patching, rotate credentials and API keys that were stored on or used by the orchestrator if you suspect prior exposure, and review recent configuration and policy changes for unauthorized edits.
• Reduce the long-term attack surface: restrict management access to dedicated admin networks or jump hosts, enforce strong authentication (including MFA where supported), and disable unused management features and interfaces.
• Ensure host and application logging is retained off-box so future incidents can be investigated.

If you can't patch immediately

Until the vendor fix is installed, reduce exposure and increase detection. Compensating controls do not replace the patch.

• Remove direct internet exposure of the orchestrator; place it behind VPN, zero-trust access, or tightly controlled jump hosts with allowlisted source IPs.
• Segment the VCO management plane from user and IoT networks; block unnecessary east-west paths from the orchestrator host.
• If a WAF or reverse proxy sits in front of the management UI, apply strict allowlists and virtual-patch rules only where they match vendor guidance—generic SQLi/XSS rules are not a substitute for command-injection fixes.
• Disable or restrict any optional features, APIs, or integrations the advisory identifies as related to the vulnerable path, if your operations allow it.
• Heighten monitoring: alert on new processes, shell activity, unexpected outbound connections, and admin authentication failures or successes from unusual locations. Prepare an isolation plan if compromise is suspected.
• If mitigations cannot be applied and risk is unacceptable, follow organizational and BOD 26-04-aligned guidance on discontinuing use until a secure configuration is available.

If your data may have been exposed

Actively exploited management-plane flaws can lead to full compromise of the orchestrator and the configurations and operational data it holds. Ransomware use is not documented for this CVE in the provided facts; still treat confirmed exploitation as a security incident: isolate affected hosts, preserve volatile and disk evidence, rotate secrets, and follow your forensics and notification procedures (including any CISA forensics triage expectations that apply to you).

If you need a quick external check on whether email addresses tied to your organization appear in known breach datasets, you can run a free exposure scan of those addresses as one input to your investigation—not as proof that this CVE was or was not used against you.