Skip to content

Traefik: Cross-user response poisoning via proxied CONNECT on Traefik's shared backend keep-alive pool

High severity GitHub Reviewed Published Jul 27, 2026 in traefik/traefik • Updated Aug 6, 2026

Package

gomod github.com/traefik/traefik (Go)

Affected versions

<= 1.7.34

Patched versions

None
gomod github.com/traefik/traefik/v2 (Go)
<= 2.11.52
2.11.53
gomod github.com/traefik/traefik/v3 (Go)
<= 3.6.23
>= 3.7.0, <= 3.7.8
3.6.24
3.7.9

Description

Summary

There is a critical vulnerability in Traefik's default HTTP reverse proxy that leads to unauthenticated cross-user response poisoning. When a client opens an HTTP/2 or HTTP/3 CONNECT request, Traefik forwards it — body included — to an HTTP/1.1 upstream over a shared net/http.Transport. If the upstream answers the CONNECT with a keep-alive non-2xx response without draining the body, the now-desynchronized backend socket is returned to Traefik's shared connection pool and reused for other clients, letting an attacker make a different client read a response the attacker smuggled — which may be another user's authenticated or private content. The entrypoint's sanitizePath option (default true) is not a reliable defense: backends that answer CONNECT / with a keep-alive non-2xx remain exploitable. The experimental FastProxy implementation was not affected. The issue is fixed by deferring the forwarded CONNECT payload until the backend accepts the tunnel, by not returning CONNECT connections to the shared idle pool, and by discarding the CONNECT body in the ForwardAuth path.

Patches

For more information

If you have any questions or comments about this advisory, please open an issue.

Original Description

Summary

Traefik's default reverse proxy forwards a plain HTTP/2 or HTTP/3 CONNECT request and its body to
an HTTP/1.1 upstream through a shared net/http.Transport. When the upstream answers the CONNECT
with a keep-alive non-2xx response and does not drain the body, Traefik returns the now
desynchronized backend socket to its shared pool and reuses it for other clients. An
unauthenticated attacker uses this to make a different client read the attacker's smuggled response.

Traefik's default proxy is net/http/httputil.ReverseProxy over a shared http.Transport, so it
inherits the same root cause as the Caddy reverse_proxy CONNECT pool poisoning.

Traefik ships one partial mitigation Caddy does not. The entrypoint option sanitizePath (default
true) rewrites the forwarded CONNECT's empty path to /, so Traefik emits CONNECT / instead of
authority-form CONNECT host:port. This is not a reliable defense. It avoids the smuggle only
against backends that reject CONNECT / by closing the connection (Apache, nginx). Backends that
answer CONNECT / with a keep-alive non-2xx and leave the body undrained still cross. That set
includes any Go net/http server and gunicorn/Flask.

Confirmed on the official image traefik:v3.6.23 (a currently supported release), default
configuration, against stock go-httpbin (Go) and kennethreitz/httpbin (Python gunicorn/Flask),
attacker and victim in separate containers, over both HTTP/2 and HTTP/3.

Affected

  • traefik:v3.6.23 (official image) and current v3, default configuration, standard proxy to an
    HTTP/1.1 upstream. Backend keep-alive pooling is on by default (MaxIdleConnsPerHost 200).
  • Attacker frontend is HTTP/2 or HTTP/3. An HTTP/1.1 frontend is not affected.
  • The upstream keeps the connection alive after a non-2xx to the forwarded CONNECT and does not
    drain the body.
  • The experimental FastProxy implementation is not affected (see Not affected).

Details

Three behaviors compose.

  1. Traefik forwards a plain CONNECT as an ordinary proxied request. The default proxy is
    httputil.ReverseProxy with a shared http.Transport (pkg/proxy/httputil/proxy.go). The
    director assigns the outbound URL.Host directly and does not reject CONNECT, leaving the
    request body a live stream. The client places a raw HTTP/1.1 request in that body (H2/H3 DATA
    frames), which is written onto the backend socket after the CONNECT header block.

  2. net/http writes the CONNECT body unframed and pools the socket. For a CONNECT the transport
    writes the body with no Content-Length and no Transfer-Encoding. The upstream answers a
    keep-alive non-2xx and parses the trailing bytes as a pipelined request. Go reads the non-2xx
    response and returns the socket to the shared idle pool once the request body reaches EOF (the
    wroteRequest gate), while the smuggled request's response is still pending.

  3. Desynchronized reuse. The smuggled request targets a slow endpoint so its response arrives after
    the socket is pooled. A different client that reuses the socket reads the pending smuggled
    response as its own.

sanitizePath (default true, pkg/server/server_entrypoint_tcp.go) calls req.URL.JoinPath(),
which turns the CONNECT's empty path into /. Traefik emits CONNECT /. Whether that stops the
smuggle depends only on the backend: Apache and nginx answer 400 Bad Request with
Connection: close (socket torn down, no cross); Go net/http and gunicorn/Flask answer a
keep-alive non-2xx and pipeline the trailing bytes (cross). With sanitizePath off, Traefik emits
authority-form CONNECT host:port, which Apache answers with a keep-alive 405.

HTTP/2 and HTTP/3 only. Pooling requires the forwarded request body to reach EOF. An H2/H3 client
half-closes the CONNECT stream (END_STREAM), so the body reaches EOF while the connection stays open
and the socket is pooled. An H1 CONNECT body is the tunnel and cannot reach EOF without closing the
connection, so the socket is closed, not pooled. HTTP/3 routes to the same handler chain as HTTPS.

Backend behavior

"Armed" means the backend answers with a keep-alive non-2xx and parses the trailing undrained bytes
as a pipelined request. Default Traefik emits CONNECT /; with sanitizePath: false it emits
authority-form CONNECT host:port.

Backend (stock image) Server CONNECT / (default) authority-form CONNECT
mccutchen/go-httpbin Go net/http armed (405 keep-alive) armed
traefik/whoami Go net/http armed (200 keep-alive) armed
caddy:2 Go net/http armed (405 keep-alive) armed
kennethreitz/httpbin gunicorn/Flask armed (405 keep-alive) armed
httpd:2.4 Apache not armed (400 close) armed (405 keep-alive)
nginx:alpine nginx not armed (400 close) not armed (400 close)
tomcat:10 Tomcat not armed (501 close) -
node http Node.js not armed (closes) -
python -m http.server Python stdlib not armed (501 close) -

Impact

Unauthenticated cross-user HTTP response poisoning. One client receives another client's response,
which can be authenticated or private content, or an attacker-chosen response.

Blast radius depends on the pool. With the default pool and a slow smuggled endpoint the crossing is
reliable for a converging victim. With a bounded pool one desync shifts the whole response queue:
measured with MaxIdleConnsPerHost 1 and a slow victim endpoint, 8 of 8 sequential victims read a
response that was not their own (1 the attacker's, 7 another user's, 0 their own). Traefik does not
expose MaxConnsPerHost, so the parallel cascade is weaker than Caddy's.

Proof of concept

poc/run.sh runs the official traefik:v3.6.23 image fronting real off-the-shelf backends over
HTTP/1.1, with attacker and victim in separate containers. Requires docker and python3. It builds
the attack client, pulls the stock images, and runs the scenarios below.

The attacker opens an H2 or H3 CONNECT to Traefik and sends a raw HTTP/1.1
GET /delay/2?tag=ATTACKERSMUGGLED as the CONNECT body, then half-closes the stream. Traefik
forwards the CONNECT to the Go/Python backend, the backend answers a keep-alive non-2xx, keeps the
socket, and parses the trailing GET as a pipelined request, so a response to it is queued on that
socket. net/http returns the socket to Traefik's shared pool. The victim then sends
GET /get?tag=VICTIMOWN on its own connection, Traefik reuses the pooled backend socket, and the
victim reads the queued /delay response instead of its own. CROSS means the victim received a
response that was not its own.

Expected output from poc

== core: DEFAULT config, cross-user poisoning vs real off-the-shelf backends ==
  [core-go-h2] h2->h2 CROSS
  [core-go-h3] h3->h3 CROSS
  [core-go-x] h2->h3 CROSS
  [core-py-h2] h2->h2 CROSS
  [core-py-h3] h3->h3 CROSS
== mechanism: sanitizePath off -> stock Apache 405 (the direct Caddy analogue) ==
  [mech-ap-h2] h2->h2 CROSS
  [mech-ap-h3] h3->h3 CROSS
== controls: must NOT cross ==
  [ctl-apache] h2->h2 NO_CROSS
  [ctl-pooloff] h2->h2 NO_CROSS
  [ctl-kaoff] h2->h2 NO_CROSS
== safe variant: experimental FastProxy chunk-frames the CONNECT body ==
  [safe-fast] h2->h2 NO_CROSS
== cascade: bounded pool, one desync poisons a queue of victims ==
  smuggled=1 other_user=7 own=0 of 8 (cross-user poisoned=8)
RESULT: PASS
  • Core rows. DEFAULT Traefik config against a Go backend (go-httpbin) and a Python gunicorn/Flask
    backend (kennethreitz/httpbin), for H2->H2, H3->H3, and H2->H3. The victim reads the attacker's
    smuggled response.
  • Mechanism rows. sanitizePath off and stock Apache. Traefik emits authority-form
    CONNECT apache-backend:80, Apache answers a keep-alive 405, and it crosses. This is the direct
    Caddy analogue and proves the full mechanism including Apache.
  • Control rows. ctl-apache runs the default config against Apache, which closes CONNECT /;
    ctl-pooloff disables Traefik backend reuse (maxIdleConnsPerHost: -1); ctl-kaoff runs Apache
    with KeepAlive Off. All three print NO_CROSS, so the crossing depends on backend socket reuse,
    not pipelining or a shared client.
  • Safe variant. Experimental FastProxy against the Go backend prints NO_CROSS because it
    chunk-frames the CONNECT body.
  • Cascade. MaxIdleConnsPerHost 1 and a slow victim endpoint. One CONNECT desync shifts the queue:
    of 8 sequential victims, 1 reads the attacker's smuggled response, 7 read another user's response,
    0 read their own.

The captured crossing (poc/evidence/RELEASE_v3.6.23_victim.json): the victim sent
GET /get?tag=VICTIM_OWN and received a 200 whose body is the response to
GET /delay/2?tag=ATTACKER_SMUGGLED with the echoed header X-Smuggled: released-v3.6.23, none of
which the victim sent.

Not affected

  • HTTP/1.1 frontend. An H1 CONNECT body cannot reach EOF without closing the connection, so the
    backend socket is not pooled.
  • Experimental FastProxy (experimental.fastProxy). It chunk-frames the forwarded CONNECT body
    (Transfer-Encoding: chunked, captured in poc/evidence/wire_fastproxy_chunked.txt), so the
    trailing bytes are read as the CONNECT body, not a pipelined request. safe-fast is NO_CROSS.

ForwardAuth

The ForwardAuth middleware with forwardBody: true and preserveRequestMethod: true re-issues the
request to the auth server as a CONNECT with the buffered body re-attached and ContentLength never
set (pkg/middlewares/auth/forward.go). The auth client writes that body unframed to the auth
server (captured on the wire), so a keep-alive non-2xx from the auth server poisons the shared
auth-client pool the same way.

Root cause

net/http pools a connection after a keep-alive non-2xx response to a CONNECT whose body it wrote
unframed. Traefik's default proxy forwards client CONNECT through a shared net/http.Transport and
applies no CONNECT rejection. sanitizePath changes the emitted request target but does not remove
the defect. Traefik's own FastProxy implementation frames the CONNECT body and does not cross, which
shows this is a property of the httputil/net/http path, not fixed by path normalization.

POC

poc.zip


References

@rtribotte rtribotte published to traefik/traefik Jul 27, 2026
Published to the GitHub Advisory Database Aug 6, 2026
Reviewed Aug 6, 2026
Last updated Aug 6, 2026

Severity

High

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v4 base metrics

Exploitability Metrics
Attack Vector Network
Attack Complexity Low
Attack Requirements Present
Privileges Required None
User interaction None
Vulnerable System Impact Metrics
Confidentiality None
Integrity None
Availability None
Subsequent System Impact Metrics
Confidentiality High
Integrity High
Availability None

CVSS v4 base metrics

Exploitability Metrics
Attack Vector: This metric reflects the context by which vulnerability exploitation is possible. This metric value (and consequently the resulting severity) will be larger the more remote (logically, and physically) an attacker can be in order to exploit the vulnerable system. The assumption is that the number of potential attackers for a vulnerability that could be exploited from across a network is larger than the number of potential attackers that could exploit a vulnerability requiring physical access to a device, and therefore warrants a greater severity.
Attack Complexity: This metric captures measurable actions that must be taken by the attacker to actively evade or circumvent existing built-in security-enhancing conditions in order to obtain a working exploit. These are conditions whose primary purpose is to increase security and/or increase exploit engineering complexity. A vulnerability exploitable without a target-specific variable has a lower complexity than a vulnerability that would require non-trivial customization. This metric is meant to capture security mechanisms utilized by the vulnerable system.
Attack Requirements: This metric captures the prerequisite deployment and execution conditions or variables of the vulnerable system that enable the attack. These differ from security-enhancing techniques/technologies (ref Attack Complexity) as the primary purpose of these conditions is not to explicitly mitigate attacks, but rather, emerge naturally as a consequence of the deployment and execution of the vulnerable system.
Privileges Required: This metric describes the level of privileges an attacker must possess prior to successfully exploiting the vulnerability. The method by which the attacker obtains privileged credentials prior to the attack (e.g., free trial accounts), is outside the scope of this metric. Generally, self-service provisioned accounts do not constitute a privilege requirement if the attacker can grant themselves privileges as part of the attack.
User interaction: This metric captures the requirement for a human user, other than the attacker, to participate in the successful compromise of the vulnerable system. This metric determines whether the vulnerability can be exploited solely at the will of the attacker, or whether a separate user (or user-initiated process) must participate in some manner.
Vulnerable System Impact Metrics
Confidentiality: This metric measures the impact to the confidentiality of the information managed by the VULNERABLE SYSTEM due to a successfully exploited vulnerability. Confidentiality refers to limiting information access and disclosure to only authorized users, as well as preventing access by, or disclosure to, unauthorized ones.
Integrity: This metric measures the impact to integrity of a successfully exploited vulnerability. Integrity refers to the trustworthiness and veracity of information. Integrity of the VULNERABLE SYSTEM is impacted when an attacker makes unauthorized modification of system data. Integrity is also impacted when a system user can repudiate critical actions taken in the context of the system (e.g. due to insufficient logging).
Availability: This metric measures the impact to the availability of the VULNERABLE SYSTEM resulting from a successfully exploited vulnerability. While the Confidentiality and Integrity impact metrics apply to the loss of confidentiality or integrity of data (e.g., information, files) used by the system, this metric refers to the loss of availability of the impacted system itself, such as a networked service (e.g., web, database, email). Since availability refers to the accessibility of information resources, attacks that consume network bandwidth, processor cycles, or disk space all impact the availability of a system.
Subsequent System Impact Metrics
Confidentiality: This metric measures the impact to the confidentiality of the information managed by the SUBSEQUENT SYSTEM due to a successfully exploited vulnerability. Confidentiality refers to limiting information access and disclosure to only authorized users, as well as preventing access by, or disclosure to, unauthorized ones.
Integrity: This metric measures the impact to integrity of a successfully exploited vulnerability. Integrity refers to the trustworthiness and veracity of information. Integrity of the SUBSEQUENT SYSTEM is impacted when an attacker makes unauthorized modification of system data. Integrity is also impacted when a system user can repudiate critical actions taken in the context of the system (e.g. due to insufficient logging).
Availability: This metric measures the impact to the availability of the SUBSEQUENT SYSTEM resulting from a successfully exploited vulnerability. While the Confidentiality and Integrity impact metrics apply to the loss of confidentiality or integrity of data (e.g., information, files) used by the system, this metric refers to the loss of availability of the impacted system itself, such as a networked service (e.g., web, database, email). Since availability refers to the accessibility of information resources, attacks that consume network bandwidth, processor cycles, or disk space all impact the availability of a system.
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

EPSS score

Exploit Prediction Scoring System (EPSS)

This score estimates the probability of this vulnerability being exploited within the next 30 days. Data provided by FIRST.
(43rd percentile)

Weaknesses

Inconsistent Interpretation of HTTP Requests ('HTTP Request/Response Smuggling')

The product acts as an intermediary HTTP agent (such as a proxy or firewall) in the data flow between two entities such as a client and server, but it does not interpret malformed HTTP requests or responses in ways that are consistent with how the messages will be processed by those entities that are at the ultimate destination. Learn more on MITRE.

CVE ID

CVE-2026-71324

GHSA ID

GHSA-3ccp-42pg-hgv6

Source code

Credits

Loading Checking history
See something to contribute? Suggest improvements for this vulnerability.