CVE-2026-7473: Arista EOS Tunnel Decapsulation Protocol Validation Bypass
Arista EOS switches with tunnel decapsulation configurations (VXLAN, GRE, or decap-groups) can be tricked into incorrectly processing and forwarding network packets that don't match their intended tunnel setup. The vulnerability stems from insufficient validation of tunnel protocol types—the switch accepts any packet destined for its decapsulation IP without confirming it uses the correct tunneling protocol. This misprocessing can lead to unexpected data forwarding and potential network manipulation. The issue is currently being exploited in the wild.
Source data · NVD / CISA · public domain
- CVSS
- 3.1 · 5.8 MEDIUM · CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:N/I:L/A:N
- Weaknesses (CWE)
- CWE-1023
- Affected products
- 102 configuration(s)
- Published / Modified
- 2026-06-05 / 2026-06-17
- KEV due date
- 2026-06-23 (added 2026-06-09)
NVD description (verbatim)
On affected platforms running Arista EOS where a tunnel decapsulation configuration—such as VXLAN (Virtual Extensible LAN), decap-groups, or a GRE (Generic Routing Encapsulation) tunnel interface—is present, the switch will incorrectly decapsulate and forward other unexpected tunneled packet with a destination IP matching its configured decapsulation IP. This occurs because the switch does not verify the tunnel protocol type, potentially leading to the unexpected processing of non-configured tunnel traffic. This issue has been reported as being exploited in the wild.
3 reference(s) · View on NVD →
SEC.co analysis · AI-assisted, reviewed against source
Technical summary
The vulnerability affects Arista EOS platforms with active tunnel decapsulation configurations. When a switch is configured to decapsulate specific tunnel types (VXLAN, GRE, or decap-groups), it validates that inbound packets match the destination IP of the decapsulation interface but fails to verify the tunnel protocol type. An attacker can send specially crafted packets with a destination IP matching the configured decapsulation endpoint but using a different or unexpected tunnel protocol, causing the switch to process and forward these packets as if they were legitimately encapsulated traffic. This represents a breakdown in the decapsulation validation chain and is categorized under CWE-1023 (Inappropriate Validation). The CVSS 3.1 score of 5.8 (MEDIUM) reflects network-adjacent attack complexity with integrity impact and cross-system scope.
Business impact
Exploitation enables network-level packet manipulation that could facilitate data theft, traffic interception, or network segmentation bypass in environments relying on VXLAN or GRE tunnel isolation for multi-tenant or security-boundary separation. Datacenter operators using Arista switches for overlay network management face potential compromise of logical network boundaries, enabling an attacker to inject or redirect traffic between isolated tenants or security zones. The fact that this is being actively exploited heightens urgency for networks where tunnel-based isolation is a critical security control.
Affected systems
Arista EOS operating system running on 40+ specific switch models, including the 7020 series (SR, SRG, TR variants), 7280 series (CR, CR2, CR3, DR3, PR3, QR variants in various port and density configurations). Any deployment of these switches with tunnel decapsulation enabled—whether VXLAN overlays, GRE tunnels, or decap-groups—is in scope. Non-affected platforms or EOS deployments without tunnel decapsulation configurations are not vulnerable.
Exploitability
The vulnerability requires network access to send specially crafted packets to the affected switch's decapsulation IP endpoint (CVSS vector: AV:N, AC:L, PR:N). No authentication or user interaction is required. An attacker positioned on the network or with Layer 3 reachability to the decapsulation interface can immediately attempt exploitation. Active in-the-wild exploitation has been reported, confirming practical exploitability and indicating that proof-of-concept or attack tools may already be circulating.
Remediation
Organizations must prioritize patching Arista EOS to a version that includes protocol type validation in the tunnel decapsulation path. Verify the specific patched EOS version against the Arista advisory and test in a non-production environment before deployment. Interim mitigations may include access control lists restricting traffic destined to tunnel decapsulation endpoints to trusted sources, or temporarily disabling non-critical tunnel decapsulation configurations pending patch availability. Network segmentation limiting access to switches from untrusted network segments can also reduce immediate risk.
Patch guidance
Obtain patched EOS versions from Arista's security advisory and release notes. Verify that the version includes fixes for tunnel protocol validation in decapsulation logic. Apply patches during scheduled maintenance windows to avoid network disruption. Test patches in a lab environment with representative tunnel configurations (VXLAN, GRE, decap-groups) to confirm correct decapsulation behavior and protocol filtering post-patch. Organizations should refer to Arista's official advisory for specific version numbers and upgrade procedures.
Detection guidance
Monitor for anomalous packets arriving at tunnel decapsulation IPs with unexpected protocol types (e.g., packets claiming to be GRE-encapsulated when only VXLAN is configured, or vice versa). Capture and analyze traffic using network TAPs or switch port mirroring to identify malformed or protocol-mismatched tunnel packets. Check EOS logs and NetFlow/sFlow exports for unusual decapsulation activity, unexpected source IPs, or forwarding decisions from tunnel interfaces that deviate from policy. Correlate timing with external threat intelligence on CVE-2026-7473 exploitation campaigns to identify potential targeted activity.
Why prioritize this
This vulnerability rates HIGH priority because: (1) it is actively exploited in the wild; (2) it affects critical network infrastructure (Arista switches) in broad deployment; (3) exploitation bypasses logical network isolation mechanisms, directly undermining multi-tenancy and security segmentation; (4) no authentication is required and network access is relatively common in datacenter or enterprise environments; (5) the impact includes cross-system scope (CVSS S:C), amplifying blast radius. Organizations using Arista EOS for overlay networks should treat this as urgent.
Risk score, explained
The CVSS 3.1 score of 5.8 (MEDIUM) reflects a network-adjacent attack with low complexity and no privilege or user-interaction requirements (AV:N, AC:L, PR:N, UI:N). The C:N rating indicates no confidentiality impact, but the I:L (integrity low) rating captures the ability to modify or redirect network traffic. The S:C (changed scope) modifier elevates risk by reflecting that the vulnerability affects resources beyond the vulnerable component (cross-subnet or cross-tenant impact). While formally MEDIUM, the active exploitation status, infrastructure criticality, and network isolation bypass capability elevate actual risk to organizations significantly above the base score.
Frequently asked questions
What is the practical difference between a correctly configured tunnel and the attack scenario described in this CVE?
In normal operation, a switch configured for VXLAN decapsulation will only process VXLAN-encapsulated packets (identifiable by protocol type or port markers). The CVE describes a situation where the switch accepts any packet with a matching destination IP, regardless of protocol type. An attacker can send a GRE-encapsulated packet (or raw packet with the decapsulation IP as destination) to a switch expecting only VXLAN, and the switch will incorrectly process it—effectively mixing tunnel types and bypassing protocol-level isolation.
Does this vulnerability require the attacker to be on the same network segment as the affected switch?
No. The CVSS vector AV:N means the vulnerability is network-adjacent or remote. An attacker needs Layer 3 (IP) reachability to the decapsulation IP endpoint, which could be an underlay network, a peering link, or any path with IP routing to the tunnel endpoint. They do not need physical access or direct LAN adjacency.
If we are not using VXLAN or GRE tunnels, are we affected?
If your Arista EOS deployments do not have any tunnel decapsulation configuration (VXLAN, GRE, or decap-groups) enabled, the vulnerable code path is not active and you are not at risk from this CVE. However, it is important to verify that no tunnel decapsulation is configured, as inadvertent or legacy configurations could activate the vulnerability.
Why is this marked as actively exploited if the CVSS score is only 5.8?
CVSS scores reflect technical severity in isolation; they do not account for real-world threat landscape, adversary activity, or organizational criticality. This CVE is being exploited in the wild because network infrastructure is a high-value target, and compromising tunnel isolation directly enables lateral movement and multi-tenancy bypass. The active exploitation status should drive prioritization regardless of the CVSS numeric score.
This analysis is provided for informational purposes and should not be construed as legal, security, or compliance advice. Organizations must verify all technical details, affected product versions, and patch availability against official Arista advisories and security bulletins. No guarantee is made regarding the completeness or applicability of this summary to your specific infrastructure. Refer to your security team, vendor documentation, and relevant incident response procedures before taking remediation actions. Active exploitation claims are based on threat intelligence available at time of publication and may change. Source: NVD (public-domain), retrieved 2026-07-15. Analysis generated by SEC.co (claude-haiku-4-5).
Related vulnerabilities
- CVE-2026-53839MEDIUMOpenClaw Hostname Validation Bypass Allows Credential Disclosure
- CVE-2026-53859MEDIUMOpenClaw Hostname Validation Bypass Before 2026.5.26
- CVE-2026-48587LOWDjango Cache Header Bypass – Whitespace Handling Vulnerability
- CVE-2026-25620MEDIUMArista NGFW Command Injection in Captive Portal (v17.4.0)
- CVE-2026-25621MEDIUMArista NGFW 17.4.0 Reports Input Validation Vulnerability
- CVE-2026-25622MEDIUMArista NGFW Captive Portal Command Injection
- CVE-2026-25623MEDIUMArista NGFW Admin Command Injection Vulnerability
- CVE-2026-25624MEDIUMArista NGFW Administrative XSS Vulnerability
Preview — this page is review (quality 1). high-value: hold for review.