HTTP Working Group | M. Nottingham |
Internet-Draft | Fastly |
Intended status: Standards Track | July 8, 2021 |
Expires: January 9, 2022 |
To aid debugging, HTTP caches often append header fields to a response explaining how they handled the request. This specification codifies that practice and updates it to align with HTTP's current caching model.¶
RFC EDITOR: please remove this section before publication ¶
Discussion of this draft takes place on the HTTP working group mailing list (ietf-http-wg@w3.org), which is archived at https://lists.w3.org/Archives/Public/ietf-http-wg/.¶
Working Group information can be found at https://httpwg.org/; source code and issues list for this draft can be found at https://github.com/httpwg/http-extensions/labels/cache-header.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as “work in progress”.¶
This Internet-Draft will expire on January 9, 2022.¶
Copyright (c) 2021 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶
To aid debugging, HTTP caches often append header fields to a response explaining how they handled the request. Unfortunately, the semantics of these headers are often unclear, and both the semantics and syntax used vary between implementations.¶
This specification defines a new HTTP response header field, "Cache-Status" for this purpose.¶
The Cache-Status HTTP response header field indicates caches' handling of the request corresponding to the response it occurs within.¶
Its value is a List ([RFC8941], Section 3.1):¶
Cache-Status = sf-list
Each member of the list represents a cache that has handled the request. The first member of the list represents the cache closest to the origin server, and the last member of the list represents the cache closest to the user (possibly including the user agent's cache itself, if it appends a value).¶
The Cache-Status header field is only applicable to responses that have been generated by an origin server. An intermediary SHOULD NOT append a Cache-Status member to responses that it generates, even if that intermediary contains a cache, except when the generated response is based upon a stored response (e.g., a 304 Not Modified or 206 Partial Content).¶
Caches determine when it is appropriate to add the Cache-Status header field to a response. Some might add it to all responses, whereas others might only do so when specifically configured to, or when the request contains a header field that activates a debugging mode.¶
When adding a value to the Cache-Status header field, caches SHOULD preserve the existing field value, to allow debugging of the entire chain of caches handling the request.¶
Each list member identifies the cache that inserted it and MUST be a String or Token. Depending on the deployment, this might be a product or service name (e.g., ExampleCache or "Example CDN"), a hostname ("cache-3.example.com"), an IP address, or a generated string.¶
Each member of the list can have parameters that describe that cache's handling of the request. While these parameters are OPTIONAL, caches are encouraged to provide as much information as possible.¶
This specification defines the following parameters:¶
hit = sf-boolean fwd = sf-token fwd-status = sf-integer ttl = sf-integer stored = sf-boolean collapsed = sf-boolean key = sf-string detail = sf-token / sf-string
"hit", when true, indicates that the request was satisfied by the cache; i.e., it was not forwarded, and the response was obtained from the cache.¶
A response that was originally produced by the origin but was modified by the cache (for example, a 304 or 206 status code) is still considered a hit, as long as it did not go forward (e.g., for validation).¶
A response that was in cache but not able to be used without going forward (e.g., because it was stale, or partial) is not considered a hit. Note that a stale response that is used without going forward (e.g., because the origin server is not available) can be considered a hit.¶
"hit" and "fwd" are exclusive; only one of them should appear on each list member.¶
"fwd" indicates that the request went forward towards the origin, and why.¶
The following parameter values are defined to explain why the request went forward, from most specific to least:¶
The most specific reason that the cache is aware of SHOULD be used. See also [I-D.ietf-httpbis-semantics], Section 4.¶
"fwd-status" indicates what status code (see [I-D.ietf-httpbis-semantics], Section 15) the next hop server returned in response to the request. Only meaningful when "fwd" is present; if "fwd-status" is not present but "fwd" is, it defaults to the status code sent in the response.¶
This parameter is useful to distinguish cases when the next hop server sends a 304 Not Modified response to a conditional request, or a 206 Partial Response because of a range request.¶
"ttl" indicates the response's remaining freshness lifetime (see [I-D.ietf-httpbis-cache], Section 4.2.1) as calculated by the cache, as an integer number of seconds, measured when the response header section is sent by the cache. This includes freshness assigned by the cache; e.g., through heuristics (see [I-D.ietf-httpbis-cache], Section 4.2.2), local configuration, or other factors. May be negative, to indicate staleness.¶
"stored" indicates whether the cache stored the response (see [I-D.ietf-httpbis-cache], Section 3); a true value indicates that it did. Only meaningful when fwd is present.¶
"collapsed" indicates whether this request was collapsed together with one or more other forward requests (see [I-D.ietf-httpbis-cache], Section 4); if true, the response was successfully reused; if not, a new request had to be made. If not present, the request was not collapsed with others. Only meaningful when fwd is present.¶
"key" conveys a representation of the cache key (see [I-D.ietf-httpbis-cache], Section 2) used for the response. Note that this may be implementation-specific.¶
"detail" allows implementations to convey additional information not captured in other parameters; for example, implementation-specific states, or other caching-related metrics.¶
For example:¶
Cache-Status: ExampleCache; hit; detail=MEMORY
The semantics of a detail parameter are always specific to the cache that sent it; even if a member of details from another cache shares the same name, it might not mean the same thing.¶
The most minimal cache hit:¶
Cache-Status: ExampleCache; hit
... but a polite cache will give some more information, e.g.:¶
Cache-Status: ExampleCache; hit; ttl=376
A stale hit just has negative freshness:¶
Cache-Status: ExampleCache; hit; ttl=-412
Whereas a complete miss is:¶
Cache-Status: ExampleCache; fwd=uri-miss
A miss that successfully validated on the back-end server:¶
Cache-Status: ExampleCache; fwd=stale; fwd-status=304
A miss that was collapsed with another request:¶
Cache-Status: ExampleCache; fwd=uri-miss; collapsed
A miss that the cache attempted to collapse, but couldn't:¶
Cache-Status: ExampleCache; fwd=uri-miss; collapsed=?0
Going through two separate layers of caching, where the cache closest to the origin responded to an earlier request with a stored response, and a second cache stored that response and later reused it to satisfy the current request:¶
Cache-Status: OriginCache; hit; ttl=1100, "CDN Company Here"; hit; ttl=545
New Cache-Status Parameters can be defined by registering them in the HTTP Cache-Status Parameters registry.¶
Registration requests are reviewed and approved by a Designated Expert, as per [RFC8126], Section 4.5. A specification document is appreciated, but not required.¶
The Expert(s) should consider the following factors when evaluating requests:¶
Registration requests should use the following template:¶
See the registry at https://iana.org/assignments/http-cache-status for details on where to send registration requests.¶
Upon publication, please create the HTTP Cache-Status Parameters registry at https://iana.org/assignments/http-cache-status and populate it with the types defined in Section 2; see Section 4 for its associated procedures.¶
Also, please create the following entry in the Hypertext Transfer Protocol (HTTP) Field Name Registry defined in [I-D.ietf-httpbis-semantics], Section 18.4:¶
Attackers can use the information in Cache-Status to probe the behaviour of the cache (and other components), and infer the activity of those using the cache. The Cache-Status header field may not create these risks on its own, but can assist attackers in exploiting them.¶
For example, knowing if a cache has stored a response can help an attacker execute a timing attack on sensitive data. Exposing the cache key can help an attacker understand modifications to the cache key, which may assist cache poisoning attacks. See [ENTANGLE] for details.¶
The underlying risks can be mitigated with a variety of techniques (e.g., use of encryption and authentication; avoiding the inclusion of attacker-controlled data in the cache key), depending on their exact nature.¶
To avoid assisting such attacks, the Cache-Status header field can be omitted, only sent when the client is authorized to receive it, or only send sensitive information (e.g., the key parameter) when the client is authorized.¶