HIGH 7.5

CVE-2026-47774: Envoy HTTP/2 Memory Exhaustion Denial of Service

Envoy, a widely-deployed HTTP/2 proxy for cloud applications, contains a memory exhaustion vulnerability that allows attackers to crash the service without authentication. The flaw stems from inadequate header size validation: cookie headers bypass size checks, and compressed header blocks can decompress to enormous sizes that aren't capped. An attacker can exploit this combination to force Envoy to allocate excessive memory, triggering out-of-memory errors and service downtime. Affected versions are older than 1.35.11, 1.36.7, 1.37.3, and 1.38.1.

Source data · NVD / CISA · public domain

CVSS
3.1 · 7.5 HIGH · CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Weaknesses (CWE)
CWE-405, CWE-409, CWE-770
Affected products
3 configuration(s)
Published / Modified
2026-06-17 / 2026-07-20

NVD description (verbatim)

Envoy is an open source edge and service proxy designed for cloud-native applications. Prior to versions 1.35.11, 1.36.7, 1.37.3, and 1.38.1, a vulnerability in Envoy's HTTP/2 downstream request processing allows an unauthenticated remote client to trigger excessive memory consumption, potentially resulting in OOM termination of the Envoy process and denial of service. The issue arises from the combination of two behaviors. First, cookie header bytes are not fully accounted for during request header size validation in Envoy. Second, HPACK header block limits in oghttp2/quiche are enforced on encoded bytes without a corresponding limit on total decoded header size. Together, these behaviors allow a malicious client to cause large decoded header allocations while bypassing the intended request header size protections. Versions 1.35.11, 1.36.7, 1.37.3, and 1.38.1 contain a fix. No complete workaround is known short of applying a fix. Possible temporary mitigations include disabling downstream HTTP/2 where operationally feasible; enforcing stricter request header and cookie limits before traffic reaches Envoy; and monitoring Envoy memory usage for abnormal growth under HTTP/2 traffic.

10 reference(s) · View on NVD →

SEC.co analysis · AI-assisted, reviewed against source

Technical summary

The vulnerability arises from two interacting weaknesses in Envoy's HTTP/2 request processing. First, cookie header bytes are not fully counted during downstream header size validation, allowing them to slip past the intended size gate. Second, oghttp2/quiche enforces HPACK decompression limits only on the encoded bytes of the header block, not on the decoded size. A malicious HTTP/2 client can craft a compressed header block that appears small when encoded but decompresses to a massive structure. By exploiting the cookie-size-validation gap alongside the decoded-size gap, attackers can trigger multi-gigabyte allocations, overwhelming Envoy's memory and causing process termination.

Business impact

Envoy instances acting as edge proxies or service mesh sidecars become denial-of-service targets. Production Kubernetes clusters using Envoy for ingress or inter-service routing face sudden pod evictions and service interruptions. Organizations relying on Envoy for API gateways or load balancing will see unpredictable crashes during peak traffic if even a single malicious client connects. Recovery requires manual intervention and restarts, during which downstream services remain unreachable. The low barrier to exploitation (no credentials, no special network conditions) and the high blast radius for cloud-native infrastructure make this a critical operational risk.

Affected systems

Directly affected: Envoy proxy versions prior to 1.35.11, 1.36.7, 1.37.3, and 1.38.1. Indirectly affected: any deployment using these Envoy versions, including Red Hat OpenShift Service Mesh installations. Organizations running Envoy as a sidecar proxy, ingress controller, or standalone gateway are at risk. The vulnerability applies to HTTP/2 downstream connections; gRPC and other HTTP/2-dependent services traversing Envoy are relevant attack vectors.

Exploitability

Exploitation is straightforward and does not require authentication, valid credentials, or special network access. An attacker needs only to establish an HTTP/2 connection to an Envoy listener and send a malicious request with a crafted header block. The attack is deterministic and repeatable, making it suitable for both directed attacks and mass scanning. No user interaction or social engineering is needed; the condition can be triggered automatically. The low complexity and network-accessible attack surface yield a CVSS score of 7.5 (HIGH).

Remediation

The definitive fix is to upgrade Envoy to a patched version: 1.35.11 or later, 1.36.7 or later, 1.37.3 or later, or 1.38.1 or later (verify exact versions against the vendor advisory for your release track). Until patching is feasible, temporary mitigations include disabling HTTP/2 on downstream listeners where operationally acceptable, deploying external request header validation (WAF or edge proxy) to reject oversized or unusual cookie headers before traffic reaches Envoy, and implementing aggressive memory alerting to catch abnormal consumption early.

Patch guidance

Prioritize upgrades to patched Envoy versions immediately, especially for public-facing or multi-tenant deployments. Test patches in non-production environments to confirm compatibility with your configuration and workload. Rolling updates are preferred to avoid service gaps; monitor memory usage and latency during the rollout. For organizations unable to patch immediately, enable strict request header limits at the load balancer or WAF layer, and set up automated restart policies for Envoy pods that exceed memory thresholds.

Detection guidance

Monitor Envoy's memory usage for anomalous growth, particularly during or after HTTP/2 ingress traffic spikes. Correlate memory increases with unusual request patterns (e.g., requests with very long or heavily compressed headers). Log and alert on any HTTP/2 connection attempts with decompressed header sizes exceeding expected thresholds (e.g., >1 MB). Envoy's stats endpoint can expose memory allocation patterns; integrate these metrics into your observability stack. Network-level detection is difficult without payload inspection, so rely primarily on Envoy's own telemetry.

Why prioritize this

HIGH severity due to (1) unauthenticated remote triggering, (2) direct denial-of-service impact on critical infrastructure, (3) low attack complexity and broad applicability to cloud deployments, and (4) lack of complete workarounds. Any organization running Envoy in production should treat this as urgent. The combination of ease of exploitation and the centrality of proxies in modern application architectures elevates this beyond a typical DoS vulnerability.

Risk score, explained

CVSS 3.1 score of 7.5 reflects HIGH severity: Attack Vector Network (AV:N), no privileges required (PR:N), no user interaction (UI:N), and direct impact to Availability (A:H). The absence of Confidentiality or Integrity impact (C:N, I:N) keeps the score below 9.0, but the Availability damage is total—targeted instances will crash and require restart.

Frequently asked questions

Can Envoy instances with HTTP/2 disabled be exploited?

No. The vulnerability is specific to HTTP/2 downstream request processing. If you have disabled HTTP/2 on all downstream listeners and use only HTTP/1.1, this CVE does not affect you. However, verify your configuration; many modern deployments enable HTTP/2 by default for performance.

Does this vulnerability affect Envoy if it is only used for upstream (service-to-service) connections?

The vulnerability is in downstream request processing, so Envoy's role as a client (upstream) is not vulnerable. However, if the same Envoy process also handles downstream ingress or has HTTP/2 listeners exposed to untrusted clients, the downstream path is at risk.

Are there any known public exploits or in-the-wild attacks?

As of the vulnerability publication date, the CVE is not listed in the CISA KEV catalog and no active exploitation is publicly confirmed. However, the simplicity of the attack and the ubiquity of Envoy in cloud infrastructure mean this should be treated as a high-priority patch regardless of current threat activity.

What is the performance impact of the temporary mitigations?

Disabling HTTP/2 will increase latency and reduce throughput for HTTP/2 clients, who will fall back to HTTP/1.1. Enforcing stricter header limits upstream may reject legitimate requests if limits are too low; coordinate with your application teams. Memory monitoring has negligible overhead but requires integration with your alerting infrastructure.

This analysis is based on publicly available vulnerability data as of the publication date. Patch versions, affected software versions, and vendor advisories should always be verified directly against the official Envoy project and Red Hat security announcements before implementing changes. Organizations should conduct their own risk assessment based on their specific deployment architecture, threat model, and business criticality. SEC.co makes no warranty regarding the completeness or timeliness of this intelligence and recommends independent validation before taking operational action. Source: NVD (public-domain), retrieved 2026-07-27. Analysis generated by SEC.co (claude-haiku-4-5).