On 18/07/11 23:36, Karl Pielorz wrote:
>
>
> --On 08 July 2011 12:47 -0300 Marcus Kool <marcus.kool_at_urlfilterdb.com>
> wrote:
>
>> Well, I still would like to know the URL because I like to observe
>> which set of URLs this eventually triggers.
>> I block with a URL filter some tracker URLs: URLs which are
>> unnecessary for showing video content but show the content provider
>> what you are doing. Sometimes these trackers cause hickups.
>
> Any URL on the BBC news site which streams video will do it - e.g.
>
> <http://www.bbc.co.uk/news/world-europe-14182135>
>
> If you're lucky you can watch it all the way through - most times it
> stops after a seemingly random amount of time, and displays the
> "twirling" buffering logo. You might also then get something like "This
> content doesn't seem to be working - try again later".
>
> Bypass squid and it works every time - right to the end.
>
> During streaming the squid access log shows lots of:
>
> "
> 1310988471.617 276 192.168.0.23 TCP_MISS/200 187 POST
> http://82.197.67.173/idle/Cinmdz02zSLOejOD/61 - DIRECT/82.197.67.173
> application/x-fcs
> 1310988472.070 453 192.168.0.23 TCP_MISS/200 187 POST
> http://82.197.67.173/idle/Cinmdz02zSLOejOD/62 - DIRECT/82.197.67.173
> application/x-fcs
> 1310988472.906 835 192.168.0.23 TCP_MISS/200 187 POST
> http://82.197.67.173/idle/Cinmdz02zSLOejOD/63 - DIRECT/82.197.67.173
> application/x-fcs
> 1310988474.008 1102 192.168.0.23 TCP_MISS/200 187 POST
> http://82.197.67.173/idle/Cinmdz02zSLOejOD/64 - DIRECT/82.197.67.173
> application/x-fcs
> "
>
> These keep running for quite a while even if the streaming has 'stopped'.
>
>
> Videos from Youtube work fine (as do most other sites) - it only appears
> to be the BBC videos that have the issue (or the BBC's iPlayer service -
> which is UK only afaik).
>
> Looking around the Web, I did find:
>
>
> <http://old.nabble.com/RTMPT-fails-randomly-using-squid-proxy-td16570049.html>
>
>
> I did go as far as commenting out the code that handled the "Suspicious
> request - double CR detected" check, but that didn't make any difference
> [in fairness I'd not seen any errors logged by squid about 'suspicious'
> headers - but that thread does seem to be related to what we're at least
> seeing].
>
> I did do some packet captures - around the time things start to play up
> I get a 'zero sized reply' error send from Squid to the client - but at
> this stage I thought I'd see if anyone else had encountered the issue...
>
>
> -Karl
double-CR and POST... hmm ... this bug?
http://bugs.squid-cache.org/show_bug.cgi?id=2829 though I notice a lack
of 4xx responses in that log trace you gave already.
Amos
-- Please be using Current Stable Squid 2.7.STABLE9 or 3.1.14 Beta testers wanted for 3.2.0.9Received on Mon Jul 18 2011 - 12:11:04 MDT
This archive was generated by hypermail 2.2.0 : Tue Jul 19 2011 - 12:00:01 MDT