[PATCH] Add --without-http-expect option

Josh Yudaken josh at smyte.com
Wed Aug 31 17:52:43 UTC 2016


I wasn't sure of any other examples of it breaking the HTTP standard,
so figured a compile time option was safer. Unfortunately bugging
Google any more than we already do probably won't get the problem
solved, and the standard "100 Continue" behavior causes 502 errors for
our end users.

For reference here is the issue on the google side:

If you still deem it useful I can update this to use an
"expect_continue"? option instead.

- josh

On Wed, Aug 31, 2016 at 10:38 AM, Maxim Dounin <mdounin at mdounin.ru> wrote:
> Hello!
> On Wed, Aug 31, 2016 at 10:08:44AM -0700, Josh Yudaken wrote:
>> # HG changeset patch
>> # User Josh Yudaken <josh at smyte.com>
>> # Date 1472606412 25200
>> #      Tue Aug 30 18:20:12 2016 -0700
>> # Node ID f81a22addc7fe1ca7486e932470756f4d73a3bb2
>> # Parent  c6372a40c2a731d8816160bf8f55a7a50050c2ac
>> Add --without-http-expect option
>> Some load balancers (specifically Google Cloud) do not support the expect
>> header. This adds an option to compile nginx without support.
> It is not clear why you are trying to add this as compile-time
> option.  The Expect header is defined by the HTTP standard, and
> compiling nginx without appropriate support looks clearly wrong.
> If Google Cloud fails to properly balance HTTP and needs special
> workarounds on nginx side, you may consider adding this as a
> run-time option instead, similar to the
> "chunked_transfer_encoding" directive, see
> http://nginx.org/r/chunked_transfer_encoding.
> Or, better yet, report the bug to Google.  It should be more or
> less trivial for them to fix this by just removing the Expect
> header from forwarded requests.
> --
> Maxim Dounin
> http://nginx.org/
> _______________________________________________
> nginx-devel mailing list
> nginx-devel at nginx.org
> http://mailman.nginx.org/mailman/listinfo/nginx-devel

More information about the nginx-devel mailing list