]> Meta Platforms, Inc.

ietf@bemasc.net

wit httpbis TCP proxying using HTTP CONNECT has long been part of the core HTTP specification. However, this proxying functionality has several important deficiencies in modern HTTP environments. This specification defines an alternative HTTP proxy service configuration for TCP connections. This configuration is described by a URI Template, similar to the CONNECT-UDP and CONNECT-IP protocols.

Introduction

History HTTP has used the CONNECT method for proxying TCP connections since HTTP/1.1. When using CONNECT, the request target specifies a host and port number, and the proxy forwards TCP payloads between the client and this destination (). To date, this is the only mechanism defined for proxying TCP over HTTP. In this specification, this is referred to as a "classic HTTP CONNECT proxy". HTTP/3 uses a UDP transport, so it cannot be forwarded using the pre-existing CONNECT mechanism. To enable forward proxying of HTTP/3, the MASQUE effort has defined proxy mechanisms that are capable of proxying UDP datagrams , and more generally IP datagrams . The destination host and port number (if applicable) are encoded into the HTTP resource path, and end-to-end datagrams are wrapped into HTTP Datagrams on the client-proxy path.

Problems HTTP clients can be configured to use proxies by selecting a proxy hostname, a port, and whether to use a security protocol. However, Classic HTTP CONNECT requests using the proxy do not carry this configuration information. Instead, they only indicate the hostname and port of the target. This prevents any HTTP server from hosting multiple distinct proxy services, as the server cannot distinguish them by path (as with distinct resources) or by origin (as in "virtual hosting"). The absence of an explicit origin for the proxy also rules out the usual defenses against server port misdirection attacks (see ) and creates ambiguity about the use of origin-scoped response header fields (e.g., "Alt-Svc" , "Strict-Transport-Security" ). Classic HTTP CONNECT requests cannot carry in-stream metadata. For example, the WRAP_UP capsule cannot be used with Classic HTTP CONNECT.

Overview This specification describes an alternative mechanism for proxying TCP in HTTP. Like and , the proxy service is identified by a URI Template. Proxy interactions reuse standard HTTP components and semantics, avoiding changes to the core HTTP protocol.

Conventions and Definitions The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

Specification A template-driven TCP transport proxy for HTTP is identified by a URI Template containing variables named "target_host" and "target_port". This URI Template and its variable values MUST meet all the same requirements as for UDP proxying (), and are subject to the same validation rules. The client MUST substitute the destination host and port number into this template to produce the request URI. The derived URI serves as the destination of a Capsule Protocol connection using the Upgrade Token "connect-tcp" (see registration in ). When using "connect-tcp", TCP payload data is sent in the payload of new Capsule Types named DATA and FINAL_DATA (see registrations in ). The ordered concatenation of these capsule payloads represents the TCP payload data. A FINAL_DATA capsule additionally indicates that the sender has closed this stream, semantically equivalent to TCP FIN. After sending a FINAL_DATA capsule, an endpoint MUST NOT send any more DATA or FINAL_DATA capsules on this data stream. (See for related requirements.) An intermediary MAY merge and split successive DATA and FINAL_DATA capsules, subject to the following requirements:

• There are no intervening capsules of other types.
• The order of payload content is preserved.
• The final emitted capsule uses the same capsule type (DATA or FINAL_DATA) as the final input capsule, and all others use the DATA capsule type.

In HTTP/1.1 In HTTP/1.1, the client uses the proxy by issuing a request as follows:

• The method SHALL be "GET".
• The request's target SHALL correspond to the URI derived from expansion of the proxy's URI Template.
• The request SHALL include a single "Host" header field containing the origin of the proxy.
• The request SHALL include a "Connection" header field with the value "Upgrade". (Note that this requirement is case-insensitive as per .)
• The request SHALL include an "Upgrade" header field with the value "connect-tcp".
• The request SHOULD include a "Capsule-Protocol: ?1" header (as recommended in ).

If the request is well-formed and permissible, the proxy MUST attempt to establish the TCP connection before sending any response status code other than "100 (Continue)" (see ). If the TCP connection is successful, the response SHALL be as follows:

• The HTTP status code SHALL be "101 (Switching Protocols)".
• The response SHALL include a "Connection" header field with the value "Upgrade".
• The response SHALL include a single "Upgrade" header field with the value "connect-tcp".
• The response SHOULD include a "Capsule-Protocol: ?1" header (as above).

If the request is malformed or impermissible, the proxy MUST return a 4XX error code. If a TCP connection was not established, the proxy MUST NOT switch protocols to "connect-tcp", and the client MAY reuse this connection for additional HTTP requests.

Templated TCP proxy example in HTTP/1.1

In HTTP/2 and HTTP/3 In HTTP/2 and HTTP/3, the proxy MUST include SETTINGS_ENABLE_CONNECT_PROTOCOL in its SETTINGS frame . The client uses the proxy by issuing an "extended CONNECT" request as follows:

• The :method pseudo-header field SHALL be "CONNECT".
• The :protocol pseudo-header field SHALL be "connect-tcp".
• The :authority pseudo-header field SHALL contain the authority of the proxy.
• The :path and :scheme pseudo-header fields SHALL contain the path and scheme of the request URI derived from the proxy's URI Template.

A templated TCP proxying request that does not conform to all of these requirements represents a client error (see ) and may be malformed (see and ). Additionally the "capsule-protocol" header field SHOULD be present with a value of "?1" (as recommended in ).

Templated TCP proxy example in HTTP/2

Use of Other Relevant Headers

Origin-scoped Headers Ordinary HTTP headers apply only to the single resource identified in the request or response. An origin-scoped HTTP header is a special response header that is intended to change the client's behavior for subsequent requests to any resource on this origin. Unlike classic HTTP CONNECT proxies, a templated TCP proxy has an unambiguous origin of its own. Origin-scoped headers apply to this origin when they are associated with a templated TCP proxy response. Here are some origin-scoped headers that could potentially be sent by a templated TCP proxy:

• "Alt-Svc"
• "Strict-Transport-Security"
• "Public-Key-Pins"
• "Accept-CH"
• "Set-Cookie" , which has configurable scope.
• "Clear-Site-Data"

Authentication Headers Authentication to a templated TCP proxy normally uses ordinary HTTP authentication via the "401 (Unauthorized)" response code, the "WWW-Authenticate" response header field, and the "Authorization" request header field (). A templated TCP proxy does not use the "407 (Proxy Authentication Required)" response code and related header fields () because they do not traverse HTTP gateways (see ). Clients SHOULD assume that all proxy resources generated by a single template share a protection space (i.e., a realm) (). For many authentication schemes, this will allow the client to avoid waiting for a "401 (Unauthorized)" response before each new connection through the proxy.

Closing Connections Connection termination is essentially symmetrical for proxies and their clients. In this section, we use the term "endpoint" to describe an implementation of this specification in either role. When closing connections, endpoints are subject to the following requirements:

• When an endpoint receives a valid TCP FIN, it MUST send a FINAL_DATA capsule.
• When an endpoint receives a valid FINAL_DATA capsule, it MUST send a TCP FIN.
• When a TCP connection reaches the TIME-WAIT or CLOSED state, the associated endpoint MUST close its send stream.
  • If the connection closed gracefully, the endpoint MUST close the send stream gracefully.
  • Otherwise, the endpoint SHOULD close the send stream abruptly, using a mechanism appropriate to the HTTP version:
    • HTTP/3: reset the stream with H3_CONNECT_ERROR; see and
    • HTTP/2: reset the stream with CONNECT_ERROR; see , Sections 6.4 and 7
    • HTTP/1.1 over TLS: TCP shutdown without a TLS closure alert; see .
    • HTTP/1.1 without TLS: TCP RST
• When the receive stream is closed abruptly or without a FINAL_DATA capsule received, the endpoint SHOULD send a TCP RST if the TCP subsystem permits it.

The mandatory behaviors above enable endpoints to detect any truncation of incoming TCP data. The recommended behaviors propagate any TCP errors through the proxy connection.

Simple graceful termination example (HTTP/3) +-------DATA{"abc"}------->+---"abc"--->| +----FIN---->+------FINAL_DATA{""}----->+----FIN---->| |<---FIN-----+<-----FINAL_DATA{""}------+<---FIN-----+ | |<----QUIC.STREAM{FIN}-----+ | +---FINACK-->+-----QUIC.STREAM{FIN}---->| | | | | | ]]>

Simple TCP RST termination example (HTTP/2) +---RST_STREAM{CON_ERR}--->+----RST---->| | | | | ]]>

Timeout example (HTTP/1.1) +-------DATA{"abc"}------->+---"abc"--->| | | (... timeout @ A ...) | | | +--FIN (no close_notify)-->+----RST---->| | | | | ]]>

RST after FIN example (HTTP/3) +-------FINAL_DATA{""}---->+-----FIN--->| | | | | | (FIN) | | (FIN) | |<---"abc"---+<----FINAL_DATA{"abc"}----+<---"abc"---+ | |<----QUIC.STREAM{FIN}-----+ | +-----RST--->+-----H3_CONNECT_ERROR---->+-----RST--->| | | | | ]]>

Additional Connection Setup Behaviors This section discusses some behaviors that are permitted or recommended in order to enhance the performance or functionality of connection setup.

Latency optimizations When using this specification in HTTP/2 or HTTP/3, clients MAY start sending TCP stream content optimistically, subject to flow control limits ( or ). Proxies MUST buffer this "optimistic" content until the TCP stream becomes writable, and discard it if the TCP connection fails. (Clients MUST NOT use "optimistic" behavior in HTTP/1.1, as this would interfere with reuse of the connection after an error response such as "401 (Unauthorized)".) Servers that host a proxy under this specification MAY offer support for TLS early data in accordance with . Clients MAY send "connect-tcp" requests in early data, and MAY include "optimistic" TCP content in early data (in HTTP/2 and HTTP/3). At the TLS layer, proxies MAY ignore, reject, or accept the early_data extension (). At the HTTP layer, proxies MAY process the request immediately, return a "425 (Too Early)" response (), or delay some or all processing of the request until the handshake completes. For example, a proxy with limited anti-replay defenses might choose to perform DNS resolution of the target_host when a request arrives in early data, but delay the TCP connection until the TLS handshake completes.

Conveying metadata This specification supports the "Expect: 100-continue" request header () in any HTTP version. The "100 (Continue)" status code confirms receipt of a request at the proxy without waiting for the proxy-destination TCP handshake to succeed or fail. This might be particularly helpful when the destination host is not responding, as TCP handshakes can hang for several minutes before failing. Clients MAY send "Expect: 100-continue", and proxies MUST respect it by returning "100 (Continue)" if the request is not immediately rejected. Proxies implementing this specification SHOULD include a "Proxy-Status" response header field in any success or failure response (i.e., status codes 101, 2XX, 4XX, or 5XX) to support advanced client behaviors and diagnostics. Clients and proxies MUST NOT send trailer fields on "connect-tcp" streams.

Applicability

Servers For server operators, template-driven TCP proxies are particularly valuable in situations where virtual-hosting is needed, or where multiple proxies must share an origin. For example, the proxy might benefit from sharing an HTTP gateway that provides DDoS defense, performs request sanitization, or enforces user authorization. The URI template can also be structured to generate high-entropy Capability URLs , so that only authorized users can discover the proxy service.

Clients Clients that support both classic HTTP CONNECT proxies and template-driven TCP proxies MAY accept both types via a single configuration string. If the configuration string can be parsed as a URI Template containing the required variables, it is a template-driven TCP proxy. Otherwise, it is presumed to represent a classic HTTP CONNECT proxy. In some cases, it is valuable to allow "connect-tcp" clients to reach "connect-tcp"-only proxies when using a legacy configuration method that cannot convey a URI Template. To support this arrangement, clients SHOULD treat certain errors during classic HTTP CONNECT as indications that the proxy might only support "connect-tcp":

• In HTTP/1.1: the response status code is "426 (Upgrade Required)", with an "Upgrade: connect-tcp" response header.
• In any HTTP version: the response status code is "501 (Not Implemented)".
  • Requires SETTINGS_ENABLE_CONNECT_PROTOCOL to have been negotiated in HTTP/2 or HTTP/3.

If the client infers that classic HTTP CONNECT is not supported, it SHOULD retry the request using the registered default template for "connect-tcp":

Registered default template

If this request succeeds, the client SHOULD record a preference for "connect-tcp" to avoid further retry delays.

Security Considerations Template-driven TCP proxying is largely subject to the same security risks as classic HTTP CONNECT. For example, any restrictions on authorized use of the proxy (see ) apply equally to both. A small additional risk is posed by the use of a URI Template parser on the client side. The template input string could be crafted to exploit any vulnerabilities in the parser implementation. Client implementers should apply their usual precautions for code that processes untrusted inputs.

Resource Exhaustion attacks A malicious client can achieve cause highly asymmetric resource usage at the proxy by colluding with a destination server and violating the ordinary rules of TCP or HTTP. Some example attacks, and mitigations that proxies can apply:

• Connection Pileup: A malicious client can attempt to open a large number of connections to exhaust the proxy's memory, port, or file descriptor limits. When using HTTP/2 or HTTP/3, each incremental TCP connection imposes a much higher cost on the proxy than on the attacker.
  • Mitigation: Limit the number of concurrent connections per client.
• Window Bloat: An attacker can grow the receive window size by simulating a "long, fat network" , then fill the window (from the sender) and stop acknowledging it (at the receiver). This leaves the proxy buffering up to 1 GiB of TCP data until some timeout, while the attacker does not have to retain a large buffer.
  • Mitigation: Limit the maximum receive window for TCP and HTTP connections, and the size of userspace buffers used for proxying. Alternatively, monitor the connections' send queues and limit the total buffered data per client.
• WAIT Abuse: An attacker can force the proxy into a TIME-WAIT, CLOSE-WAIT, or FIN-WAIT state until the timer expires, tying up a proxy-to-destination 4-tuple for up to four minutes after the client's connection is closed.
  • Mitigation: Limit the number of connections for each client to each destination, even if those connections are in a waiting state and the corresponding CONNECT stream is closed. Alternatively, allocate a large range of IP addresses for TCP connections (especially in IPv6).

Operational Considerations

Avoiding HTTP/1.1 While this specification is fully functional under HTTP/1.1, performance-sensitive deployments SHOULD use HTTP/2 or HTTP/3 instead. When using HTTP/1.1:

• Each CONNECT request requires a new TCP and TLS connection, imposing a higher cost in setup latency, congestion control convergence, CPU time, and data transfer.
• The graceful and abrupt closure signals () are more likely to be missing or corrupted:
  • Some implementations may be unable to emit the recommended abrupt closure signals, due to limitations in their TCP and TLS subsystems.
  • Faulty implementations may fail to send a TLS closure alert during graceful shutdown, or fail to report an error when the expected closure alert is not received. These misbehaviors are not compliant with , but they are common nonetheless among HTTP/1.1 implementations today.
• The number of active connections through each client may be limited by the number of available TCP client ports, especially if:
  • The client only has one IP address that can be used to reach the proxy.
  • The client is shared between many parties, such as when acting as a gateway or concentrator.
  • The proxied connections are often closed by the destination. This causes the client to initiate closure of the client-to-proxy connection, leaving the client in a TIME-WAIT state for up to four minutes.

Gateway Compatibility Templated TCP proxies can make use of standard HTTP gateways and path-routing to ease implementation and allow use of shared infrastructure. However, current gateways might need modifications to support TCP proxy services. To be compatible, a gateway must:

• support Extended CONNECT (if acting as an HTTP/2 or HTTP/3 server).
• support HTTP/1.1 Upgrade to "connect-tcp" (if acting as an HTTP/1.1 server)
  • only after forwarding the upgrade request to the origin and observing a success response.
• forward the "connect-tcp" protocol to the origin.
• convert "connect-tcp" requests between all supported HTTP server and client versions.
• allow any "Proxy-Status" headers to traverse the gateway.

IANA Considerations

New Upgrade Token IF APPROVED, IANA is requested to add the following entry to the HTTP Upgrade Token Registry:

Value	Description	Reference
"connect-tcp"	Proxying of TCP payloads	(This document)

For interoperability testing of this draft version, implementations SHALL use the value "connect-tcp-07".

New MASQUE Default Template IF APPROVED, IANA is requested to add the following entry to the "MASQUE URI Suffixes" registry:

Path Segment	Description	Reference
tcp	TCP Proxying	(This document)

New Capsule Type IF APPROVED, IANA is requested to add the following entry to the "HTTP Capsule Types" registry:

Value	Capsule Type	Status	Reference	Change Controller	Contact
(TBD)	DATA	permanent	(This document),	IETF	HTTPBIS
(TBD)	FINAL_DATA	permanent	(This document),	IETF	HTTPBIS

For this draft version of the protocol, the Capsule Type values 0x2028d7f0 and 0x2028d7f1 shall be used provisionally for testing, under the names "DATA-08" and "FINAL_DATA-08".

References Normative References The Hypertext Transfer Protocol (HTTP) is a stateless application-level protocol for distributed, collaborative, hypertext information systems. This document describes the overall architecture of HTTP, establishes common terminology, and defines aspects of the protocol that are shared by all versions. In this definition are core protocol elements, extensibility mechanisms, and the "http" and "https" Uniform Resource Identifier (URI) schemes. This document updates RFC 3864 and obsoletes RFCs 2818, 7231, 7232, 7233, 7235, 7538, 7615, 7694, and portions of 7230. This document describes HTTP Datagrams, a convention for conveying multiplexed, potentially unreliable datagrams inside an HTTP connection. In HTTP/3, HTTP Datagrams can be sent unreliably using the QUIC DATAGRAM extension. When the QUIC DATAGRAM frame is unavailable or undesirable, HTTP Datagrams can be sent using the Capsule Protocol, which is a more general convention for conveying data in HTTP connections. HTTP Datagrams and the Capsule Protocol are intended for use by HTTP extensions, not applications. In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. A URI Template is a compact sequence of characters for describing a range of Uniform Resource Identifiers through variable expansion. This specification defines the URI Template syntax and the process for expanding a URI Template into a URI reference, along with guidelines for the use of URI Templates on the Internet. [STANDARDS-TRACK] This document describes how to proxy UDP in HTTP, similar to how the HTTP CONNECT method allows proxying TCP in HTTP. More specifically, this document defines a protocol that allows an HTTP client to create a tunnel for UDP communications through an HTTP server that acts as a proxy. This document defines a mechanism for running the WebSocket Protocol (RFC 6455) over a single stream of an HTTP/2 connection. The mechanism for running the WebSocket Protocol over a single stream of an HTTP/2 connection is equally applicable to HTTP/3, but the HTTP-version-specific details need to be specified. This document describes how the mechanism is adapted for HTTP/3. This specification describes an optimized expression of the semantics of the Hypertext Transfer Protocol (HTTP), referred to as HTTP version 2 (HTTP/2). HTTP/2 enables a more efficient use of network resources and a reduced latency by introducing field compression and allowing multiple concurrent exchanges on the same connection. This document obsoletes RFCs 7540 and 8740. The QUIC transport protocol has several features that are desirable in a transport for HTTP, such as stream multiplexing, per-stream flow control, and low-latency connection establishment. This document describes a mapping of HTTP semantics over QUIC. This document also identifies HTTP/2 features that are subsumed by QUIC and describes how HTTP/2 extensions can be ported to HTTP/3. This document defines the core of the QUIC transport protocol. QUIC provides applications with flow-controlled streams for structured communication, low-latency connection establishment, and network path migration. QUIC includes security measures that ensure confidentiality, integrity, and availability in a range of deployment circumstances. Accompanying documents describe the integration of TLS for key negotiation, loss detection, and an exemplary congestion control algorithm. This document specifies version 1.3 of the Transport Layer Security (TLS) protocol. TLS allows client/server applications to communicate over the Internet in a way that is designed to prevent eavesdropping, tampering, and message forgery. This document updates RFCs 5705 and 6066, and obsoletes RFCs 5077, 5246, and 6961. This document also specifies new requirements for TLS 1.2 implementations. Using TLS early data creates an exposure to the possibility of a replay attack. This document defines mechanisms that allow clients to communicate with servers about HTTP requests that are sent in early data. Techniques are described that use these mechanisms to mitigate the risk of replay. This document defines the Proxy-Status HTTP response field to convey the details of an intermediary's response handling, including generated errors. Informative References This document describes how to proxy UDP in HTTP, similar to how the HTTP CONNECT method allows proxying TCP in HTTP. More specifically, this document defines a protocol that allows an HTTP client to create a tunnel for UDP communications through an HTTP server that acts as a proxy. This document describes how to proxy IP packets in HTTP. This protocol is similar to UDP proxying in HTTP but allows transmitting arbitrary IP packets. More specifically, this document defines a protocol that allows an HTTP client to create an IP tunnel through an HTTP server that acts as an IP proxy. This document updates RFC 9298. This document specifies "Alternative Services" for HTTP, which allow an origin's resources to be authoritatively available at a separate network location, possibly accessed with a different protocol configuration. This specification defines a mechanism enabling web sites to declare themselves accessible only via secure connections and/or for users to be able to direct their user agent(s) to interact with given sites only over secure connections. This overall policy is referred to as HTTP Strict Transport Security (HSTS). The policy is declared by web sites via the Strict-Transport-Security HTTP response header field and/or by other means, such as user agent configuration, for example. [STANDARDS-TRACK] Google LLC Cloudflare HTTP intermediaries sometimes need to terminate long-lived request streams in order to facilitate load balancing or impose data limits. However, Web browsers commonly cannot retry failed proxied requests when they cannot ascertain whether an in-progress request was acted on. To avoid user-visible failures, it is best for the intermediary to inform the client of upcoming request stream terminations in advance of the actual termination so that the client can wrap up existing operations related to that stream and start sending new work to a different stream or connection. This document specifies a new "WRAP_UP" capsule that allows a proxy to instruct a client that it should not start new requests on a tunneled connection, while still allowing it to finish existing requests. This document defines a new HTTP header that allows web host operators to instruct user agents to remember ("pin") the hosts' cryptographic identities over a period of time. During that time, user agents (UAs) will require that the host presents a certificate chain including at least one Subject Public Key Info structure whose fingerprint matches one of the pinned fingerprints for that host. By effectively reducing the number of trusted authorities who can authenticate the domain during the lifetime of the pin, pinning may reduce the incidence of man-in-the-middle attacks due to compromised Certification Authorities. HTTP defines proactive content negotiation to allow servers to select the appropriate response for a given request, based upon the user agent's characteristics, as expressed in request headers. In practice, user agents are often unwilling to send those request headers, because it is not clear whether they will be used, and sending them impacts both performance and privacy. This document defines an Accept-CH response header that servers can use to advertise their use of request headers for proactive content negotiation, along with a set of guidelines for the creation of such headers, colloquially known as "Client Hints." This document defines the HTTP Cookie and Set-Cookie header fields. These header fields can be used by HTTP servers to store state (called cookies) at HTTP user agents, letting the servers maintain a stateful session over the mostly stateless HTTP protocol. Although cookies have many historical infelicities that degrade their security and privacy, the Cookie and Set-Cookie header fields are widely used on the Internet. This document obsoletes RFC 2965. [STANDARDS-TRACK] This document specifies a set of TCP extensions to improve performance over paths with a large bandwidth * delay product and to provide reliable operation over very high-speed paths. It defines the TCP Window Scale (WS) option and the TCP Timestamps (TS) option and their semantics. The Window Scale option is used to support larger receive windows, while the Timestamps option can be used for at least two distinct mechanisms, Protection Against Wrapped Sequences (PAWS) and Round-Trip Time Measurement (RTTM), that are also described herein. This document obsoletes RFC 1323 and describes changes from it.

Acknowledgments Thanks to Amos Jeffries, Tommy Pauly, Kyle Nekritz, David Schinazi, and Kazuho Oku for close review and suggested changes.