Re: Clarification on HTTP caching (RFC 7234)

So to clarify on the P.S:

must-revalidate specifies the behavior as "cannot reach" in
https://tools.ietf.org/html/rfc7234#section-5.2.2.1 but does not use the
word "disconnect"

"In all circumstances a cache MUST obey the must-revalidate directive; in
particular, if a cache cannot reach the origin server for any reason, it
MUST generate a 504 (Gateway Timeout) response."

There's a reference to "disconnected" in
https://tools.ietf.org/html/rfc7234#section-4.2.4 with similar language
that seems to indicate that 5.2.2.1 is also a "disconnect"

A cache MUST NOT send stale responses unless it is disconnected (i.e., it
cannot contact the origin server or otherwise find a forward path) or doing
so is explicitly allowed (e.g., by the max-stale request directive; see
Section 5.2.1).

"Likewise, a cache SHOULD generate a 112 warn-code (see Section 5.5.3) in
stale responses if the cache is disconnected."

The confusion comes from https://tools.ietf.org/html/rfc7234#section-5.5.3
which says that

"A cache SHOULD generate this if it is intentionally disconnected from the
rest of the network for a period of time."

It's that "intentionally" that is confusing -- that seems to indicate that
the cache has an "offline" mode that can be used when it does not want to
revalidate results.  Also, what happens when a cache is unintentionally
disconnected from the rest of the netework?

So, are these all talking about the same thing?

Will.



On Tue, Apr 28, 2015 at 11:37 AM, Will Sargent <will.sargent@gmail.com>
wrote:

> Hi all,
>
> I have some questions on the behavior of RFC 7234's s-maxage response
> directive -- it looks like the behavior for s-maxage response directive
> differs significantly from maxage in that a max-age response directive
> means that a stale response calls for "validate", where the behavior is
> "validate, but serve stale on a timeout or (if stale-if-error is defined)
> on a 5xx".
>
> Meanwhile, s-maxage says " The s-maxage directive also implies the
> semantics of the proxy-revalidate response directive."  Well,
> proxy-revalidate is the same as must-revalidate, but only operating on
> shared caches, and the validation behavior there is "validate or 503 on
> timeout".
>
> So the question is: the behavior of s-maxage explicitly to timeout rather
> than give a stale response?  I've been looking around the internets, and I
> don't see this behavior mentioned in the blogs or even in the O'Reilly Web
> Caching book or HTTP The Definitive Guide.
>
> Will.
>
> P.S.  I'm also confused as to what the difference between a "disconnect'
> and a "timeout" is -- in one area, a disconnect is referred to as "cannot
> reach the origin server" and in the warning it's referred to as
> "intentionally disconnected", but there's no direct reference to it.
>