CVEReports
CVEReports

Automated vulnerability intelligence platform. Comprehensive reports for high-severity CVEs generated by AI.

Product

  • Home
  • Sitemap
  • RSS Feed

Company

  • About
  • Contact
  • Privacy Policy
  • Terms of Service

© 2026 CVEReports. All rights reserved.

Made with love by Amit Schendel & Alon Barad



CVE-2026-71324

CVE-2026-71324: Cross-User Response Poisoning in Traefik via HTTP/2 and HTTP/3 CONNECT Handling

Amit Schendel
Amit Schendel
Senior Security Researcher

Aug 6, 2026·7 min read·80 visits

Executive Summary (TL;DR)

An unauthenticated remote attacker can poison the backend connection pool in Traefik by sending HTTP/2 or HTTP/3 CONNECT requests containing smuggled payloads. When the upstream HTTP/1.1 server rejects the tunnel, unconsumed payload bytes remain in the connection buffer and are served to other users.

CVE-2026-71324 is a high-severity HTTP request smuggling vulnerability in the Traefik reverse proxy. It allows an unauthenticated remote attacker to achieve cross-user response poisoning when Traefik is configured to route HTTP/2 or HTTP/3 CONNECT requests to an HTTP/1.1 upstream backend. By sending a crafted CONNECT request that is subsequently rejected by the backend with a keep-alive non-2xx response, the attacker can leave smuggled requests within the shared connection pool, which are then served to subsequent clients.

Vulnerability Overview

Traefik is a widely deployed, open-source HTTP reverse proxy and load balancer written in Go. The software acts as an entry point for microservices, routing external traffic to internal services using various protocols. The vulnerability exists in the transition layer where incoming HTTP/2 or HTTP/3 protocols are mapped down to HTTP/1.1 connections for communication with backend servers.

Under modern HTTP specifications, specifically RFC 8441, clients can initiate a raw TCP tunnel over multiplexed HTTP/2 or HTTP/3 streams using the CONNECT method. This mechanism allows protocols like WebSockets or custom TCP-based streams to share a single, multiplexed transport channel. When Traefik processes these incoming streams, it establishes a corresponding HTTP/1.1 connection to the upstream server and forwards the headers along with the payload.

The core of the vulnerability lies in the lack of correlation between the lifecycle of the downstream tunnel stream and the upstream HTTP/1.1 connection. When the backend rejects the connection request but leaves the connection open, Traefik returns the socket to the connection pool while client-controlled bytes are still pending. This mismatch creates an attack surface where subsequent client requests get associated with smuggled data.

Root Cause Analysis

The root cause of CVE-2026-71324 is a desynchronization between Go's internal net/http client transport and the state of physical TCP sockets within Traefik's idle connection pool. When a client issues an HTTP/2 or HTTP/3 CONNECT request, Traefik forwards this request over an HTTP/1.1 socket managed by Go's net/http.Transport. If the backend processes the header and decides to reject the tunnel, it responds with a status code such as 401 Unauthorized or 405 Method Not Allowed.

Upon receiving this non-2xx response, the backend does not read or drain any data from the request body because it has already determined that the transaction has failed. Crucially, the backend leaves the connection open by supplying a keep-alive header. When the net/http.Transport reads the non-2xx response, it determines that the HTTP transaction is finished and returns the backend TCP socket directly to the shared idle connection pool.

However, the client's tunnel data has already been written to that same socket or is still being transmitted. Because the backend rejected the request without reading the remaining request body, those unconsumed bytes reside within the socket's receive buffer on the backend. This unconsumed payload contains complete, syntactically valid HTTP/1.1 requests drafted by the attacker. When another user's request is dispatched over this recycled connection, the backend processes the attacker's smuggled request first, leading to response poisoning.

Code Analysis

The vulnerability has been addressed across three separate pull requests. The first patch modifies the ForwardAuth middleware to prevent tunnel data from being sent to the authentication server when the length is unknown.

// pkg/middlewares/auth/forward.go
 
	forwardBody := fa.forwardBody
	// When a CONNECT method has a body with an unknown length we consider the bytes as tunnel data.
	// Therefore, we do not want to forward them to the auth server.
	if req.Method == http.MethodConnect && req.ContentLength < 0 {
		forwardBody = false
	}
 
	if forwardBody {
		forwardReq.ContentLength = req.ContentLength
		forwardReq.TransferEncoding = req.TransferEncoding
 
		bodyBytes, err := fa.readBodyBytes(req)

The second patch targets the core proxy implementation in pkg/server/service/proxy.go. It forces the proxy to set Close = true on outgoing CONNECT requests, ensuring the underlying socket is closed after the transaction and never returned to the pool.

// pkg/server/service/proxy.go
 
			// Adding the "Connection: close" header to the request ensures that we are not reusing the connection for
			// subsequent requests in case the backend does not support CONNECT and returns a 2xx response.
			if pr.Out.Method == http.MethodConnect {
				pr.Out.Close = true
			}

The third patch introduces a new handler, connectHandler, which acts as a deferral mechanism for the payload. It blocks delivery of the CONNECT body until the upstream server explicitly returns a successful 2xx status code.

// pkg/server/service/connect.go
 
func (h *connectHandler) ServeHTTP(rw http.ResponseWriter, req *http.Request) {
	if req.Method != http.MethodConnect {
		h.next.ServeHTTP(rw, req)
		return
	}
	// ...
	pipeReader, pipeWriter := io.Pipe()
	tunnel := &connectTunnel{in: pipeWriter, data: req.Body}
	// The Transport blocks on the empty pipe, preventing the body from reaching the backend
	req.Body = pipeReader
	h.next.ServeHTTP(&connectResponseWriter{ResponseWriter: rw, req: req, tunnel: tunnel}, req)
}

In the custom connectResponseWriter, if the status code is not 2xx, the tunnel is closed and the buffered payload bytes are discarded without being written to the socket.

Exploitation Methodology

To exploit this vulnerability, an attacker must identify a path where Traefik handles HTTP/2 or HTTP/3 traffic and forwards it to an HTTP/1.1 backend. The target endpoint must also be configured to reject the CONNECT method or return a keep-alive non-2xx status code. The attacker then structures an HTTP/2 CONNECT request where the payload is configured to look like a standard HTTP/1.1 GET or POST request.

Because the backend processes the buffered bytes first, it generates a response to the smuggled request first. The proxy maps this response to the first available active request on the connection, which belongs to the victim client. The victim then receives the response intended for the attacker's smuggled request, potentially exposing sensitive data or executing unauthorized scripts in the context of the victim's session.

Impact Assessment

The security impact of CVE-2026-71324 is classified as high. Successful exploitation leads to cross-user response poisoning, which breaks the fundamental isolation between different clients sharing the same proxy. An attacker can manipulate what a victim receives, potentially injecting malicious JavaScript to hijack accounts or redirect traffic.

The CVSS v4.0 base score is calculated as 7.0 (High). The vulnerability does not require authentication and has low attack complexity, although it requires specific preconditions such as an upstream backend that rejects CONNECT requests while maintaining keep-alive. The primary impact is classified under subsequent confidentiality and integrity, as arbitrary responses can be forced upon unrelated sessions.

In scenarios where the ForwardAuth middleware is active with forwardBody enabled, the auth pool is also exposed. An attacker could poison the authentication connection pool, causing subsequent users' credentials or sessions to be processed against incorrect context, or potentially bypassing authentication checks entirely.

Remediation & Mitigation Guidance

The primary recommendation is to upgrade Traefik instances to one of the patched releases: 2.11.53, 3.6.24, or 3.7.9. These updates contain the deferred body handling logic and connection closure configurations that prevent socket reuse after a rejected CONNECT transaction.

If patching is not immediately feasible, administrators can apply several mitigation strategies. Disabling HTTP/2 and HTTP/3 support on Traefik entrypoints prevents the processing of multiplexed CONNECT streams, which completely disables the primary attack vector. This can be configured by modifying the entrypoints section of the Traefik configuration file to only negotiate HTTP/1.1.

Additionally, if the ForwardAuth middleware is in use, verify that forwardBody and preserveRequestMethod are set to false. Deploying Web Application Firewall (WAF) rules to detect and drop incoming requests utilizing the CONNECT method can also prevent exploit attempts from reaching the proxy. These mitigations should be applied incrementally until a formal software upgrade is completed.

Official Patches

Traefik LabsPull request addressing CONNECT state deferral.
Traefik LabsPull request correcting ForwardAuth and FastProxy state logic.
Traefik LabsPull request regulating connection closure behavior for CONNECT requests.

Fix Analysis (3)

Technical Appendix

CVSS Score
7.0/ 10
CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:N/VI:N/VA:N/SC:H/SI:H/SA:N

Affected Systems

Traefik Proxy (versions < 2.11.53)Traefik Proxy (versions 3.0.0 to 3.6.23)Traefik Proxy (versions 3.7.0 to 3.7.8)

Affected Versions Detail

Product
Affected Versions
Fixed Version
Traefik
Traefik Labs
< 2.11.532.11.53
Traefik
Traefik Labs
>= 3.0.0, < 3.6.243.6.24
Traefik
Traefik Labs
>= 3.7.0, < 3.7.93.7.9
AttributeDetail
CWE IDCWE-444
Attack VectorNetwork
CVSS v4.0 Score7.0 (High)
Exploit StatusProof-of-Concept
KEV StatusNot Listed
ImpactCross-User Response Poisoning

MITRE ATT&CK Mapping

T1557.002Adversary-in-the-Middle: HTTP Request Smuggling
Credential Access
CWE-444
Inconsistent Interpretation of HTTP Requests ('HTTP Request/Response Smuggling')

The product does not properly parse, interpret, or handle inconsistent or malformed HTTP requests, which may allow attackers to pass smuggled headers or commands across trust boundaries.

Known Exploits & Detection

GitHub Security AdvisoryOriginal security disclosure and analysis document.

Vulnerability Timeline

Patches drafted for body deferral and authorization corrections
2026-07-22
Patches drafted for connection pool closure limits
2026-07-23
Coordinated vulnerability advisory GHSA-3ccp-42pg-hgv6 published
2026-08-06
CVE-2026-71324 registered in the global CVE index
2026-08-06

References & Sources

  • [1]GHSA-3ccp-42pg-hgv6: Response poisoning via CONNECT requests
  • [2]CVE.org Authority Record for CVE-2026-71324
  • [3]Traefik v3.7.9 Release Notes

Attack Flow Diagram

Press enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

More Reports

•2 days ago•CVE-2026-16454
4.3

CVE-2026-16454: Object-Level Authorization Bypass in Eclipse hawkBit Direct Device Integration Controller

CVE-2026-16454 is an object-level authorization bypass vulnerability (CWE-284, CWE-862) in Eclipse hawkBit versions 1.0.3 and prior. The flaw exists within the Direct Device Integration (DDI) REST controller (hawkbit-ddi-resource), where endpoints responsible for enumerating software module metadata and serving firmware binaries fail to verify whether the requested software module is assigned to the authenticated device. Consequently, any device with valid credentials (such as a TargetToken or GatewayToken) can enumerate software module identifiers and download arbitrary firmware artifacts across the entire tenant namespace.

Amit Schendel
Amit Schendel
10 views•9 min read
•2 days ago•CVE-2026-62284
8.6

CVE-2026-62284: Remote Code Execution and SSRF via Unsanitized Kubeconfig in OSAC Operator

CVE-2026-62284 (GHSA-5w5g-j6hw-6p8f) is a code injection and server-side request forgery vulnerability in the console-proxy component of the OSAC Operator. The RemoteConfigResolver parses Kubernetes Secrets containing kubeconfig files via clientcmd.RESTConfigFromKubeConfig without sanitizing execution-related fields. An attacker with RBAC permissions to create or modify labeled Secrets in monitored namespaces can supply a malicious exec credential plugin, resulting in arbitrary command execution within the console-proxy pod.

Amit Schendel
Amit Schendel
8 views•8 min read
•2 days ago•CVE-2026-62361
5.5

CVE-2026-62361: Authenticated SQL Injection and Table Access Control Bypass in Listmonk Subscriber Export

CVE-2026-62361 (GHSA-xgjr-7j9q-2h4r) is an authenticated SQL injection and authorization bypass vulnerability in listmonk, an open-source, self-hosted newsletter and mailing list manager written in Go. In versions prior to 6.2.0, the subscriber export endpoint (GET /api/subscribers/export) directly interpolates user-supplied SQL filter expressions into a PostgreSQL query without invoking the table allowlist validation function (validateQueryTables). Consequently, authenticated users possessing the subscribers:sql_query and subscribers:get_all permissions can bypass relational access controls, execute arbitrary subqueries against restricted tables such as users and settings, and modify database state via writable Common Table Expressions (CTEs).

Alon Barad
Alon Barad
9 views•11 min read
•2 days ago•CVE-2026-28385
5.0

CVE-2026-28385: Server-Side Request Forgery in Canonical LXD Image Import

CVE-2026-28385 (GHSA-3gq2-x4qg-p4g6) is a Server-Side Request Forgery (SSRF) vulnerability in Canonical LXD versions 4.12 through 6.9. An authenticated user possessing the can_create_images entitlement can force the LXD daemon to execute arbitrary outbound HTTP and HTTPS requests via the /1.0/images API endpoint. Due to the absence of egress destination validation, an attacker can probe host loopback services, RFC 1918 private network segments, and cloud Instance Metadata Services (IMDS) from the network position of the privileged LXD daemon.

Amit Schendel
Amit Schendel
10 views•9 min read
•2 days ago•GHSA-9J25-MM2H-2F76
9.9

CVE-2026-48750: Arbitrary File Write on Host via exec-output Symlink in Canonical LXD and Incus

Canonical LXD and LXC Incus contain an arbitrary file write vulnerability (CWE-73 / CWE-59) in the `/1.0/instances/<name>/exec` REST API endpoint. When an instance is unpacked from an image archive, top-level symbolic links are extracted directly onto the host filesystem within the instance storage directory. If a crafted image includes a symbolic link named `exec-output` pointing to an arbitrary host path, invoking the execution endpoint with `record-output: true` causes the privileged host daemon to follow the symlink and write command output files directly to the target host directory. An authenticated attacker with permissions to import images and execute commands can exploit this flaw to write controlled content to sensitive host directories such as `/etc/cron.d`, resulting in arbitrary code execution on the host as `root`.

Alon Barad
Alon Barad
8 views•12 min read
•2 days ago•GHSA-VGHH-5RFX-XHQ8
9.9

CVE-2026-48749: Arbitrary Host File Read and Write via Malicious Image Symlink in Canonical LXD and Incus

CVE-2026-48749 (tracked as GHSA-vghh-5rfx-xhq8 for LXD and GHSA-2q3f-q5pq-g8wv for Incus) is a critical container escape vulnerability in the Canonical LXD and Linux Containers Incus system container managers. By importing a specially crafted container image archive containing duplicate tar member headers, an authenticated attacker can replace the instance rootfs directory with a symbolic link pointing to the underlying host root filesystem. When performing file operations on a stopped container, the privileged forkfile helper binary resolves the symbolic link without confinement and chroots directly into the host filesystem, granting arbitrary root-level file read and write access.

Amit Schendel
Amit Schendel
8 views•9 min read