Re: 2.6.17.1: fails to fully get webpage

From: CaT <cat@xxxxxxxxxx>
Date: Thu, 29 Jun 2006 13:09:23 +1000

Yup. Must've. Just tried it now and setting tcp_window_scaling to 0
makes the site load. Sorry about that. Looks like I'll have to remember
to make sure that gets set to zero on any servers I wind up having to
upgrade to 2.6.17 and beyond.

Bugger. :/

You can save yourself that hassle by informing the site admin
of the affected site that they have a firewall that misinterprets
the RFC standard window scaling field of the TCP headers. These
devices assume it is zero because they don't remember the window
scale negotiated at the beginning of the TCP connection.

Your TCP performance will suffer greatly if you disable window
scaling across the board. It means that only a 64K window will
be usable by TCP, and you'll not be able to fill the pipe.

Please don't use a screwdriver to pound in a nail :)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • Re: tcp messages
    ... The mechanisms provided allow a TCP to advertise a large window and to ... The sending TCP must be prepared to accept from the user and send at least ... one octet of new data even if the send window is zero. ... Two minutes is recommended for the retransmission ...
    (RedHat)
  • Re: fetch hangs on AMD64 RELENG_6
    ... retransmit when we open it. ... sends a packet opening the window. ... "The sending TCP must be prepared to accept from the user and send at ... least one octet of new data even if the send window is zero. ...
    (freebsd-stable)
  • Re: fetch hangs on AMD64 RELENG_6
    ... retransmit when we open it. ... Normally the stack only sets the window size to 0 in the event of severe congestion, it's used to tell the other side to stop sending traffic for an interval, although the other side should retry with zero-data-length ACK-only packets after a delay, or once your side sends a packet opening the window. ... "The sending TCP must be prepared to accept from the user and send at ... least one octet of new data even if the send window is zero. ...
    (freebsd-stable)
  • Re: Interesting TCP issue
    ... }} Subject: Re: Interesting TCP issue ... }}> idiotically small window the remote server is advertising. ... }} turn off window scaling and see if that helps ...
    (freebsd-hackers)
  • RE: RWIN (TCP Window Size) negotiates properly after upgrading to SP2
    ... Apparently SP2 uses a different registry setting than SP1: ... TCP window size. ... There are 3-4 places in the Windows Registry that affect the TCP Window. ...
    (microsoft.public.windowsxp.network_web)