Re: [squid-users] Squid 3.1.1 and flash video scrubbing

From: Amos Jeffries <squid3_at_treenet.co.nz>
Date: Fri, 09 Apr 2010 12:29:49 +1200

Henrik Nordström wrote:
> tor 2010-04-08 klockan 20:43 +1200 skrev Amos Jeffries:
>
>> FWIW, the flash player and the server are somewhat broken and playing
>> bad games with HTTP/1.1 Range requests.
>
> Which quite likely triggers confusion within Squid as the store/object
> interface is range aware. Has been issues there in the past, and quite
> likely is more issues..
>
> Not 100% sure that response is technically broken, but it surely is an
> odd response.
>
>> It claims to be cacheable but isn't.
>
> Why isn't it?

I though Squid considered it a malformed 206 and uncachabled it.

Does squid consider the reply a malformed 206 or a malformed 200?

The extra bytes which exist in the body means we should be assuming its
a malformed 200 with unusable range headers.

>
>> If this "range" was merged into a
>> previous ranges of the object, or even fetched from the a full copy of
>> real object by any well behaved middleware proxy it would corrupt the
>> media transfer.
>
> We don't know how the server would react on requests for the same URI
> but with Range header..

Server may or may not provide a real range or this same output.

Regardless of that any client being smart and fetching the request as a
range of the indicated range bytes from a full copy of the object will
get different bytes from any intermediary than this reply contains.

>
> There is no Content-Location header in the response, so merging of
> responses with different url parameters MUST NOT be done by any HTTP
> agent (including proxies). Content-aware merging of the data is allowed,
> but that's outside HTTP.

Ah. Thanks.

Amos

-- 
Please be using
   Current Stable Squid 2.7.STABLE9 or 3.1.1
Received on Fri Apr 09 2010 - 00:30:01 MDT

This archive was generated by hypermail 2.2.0 : Fri Apr 09 2010 - 12:00:03 MDT