https://bz.apache.org/bugzilla/show_bug.cgi?id=56188

--- Comment #9 from Luca Toscano <[email protected]> ---
So while reading http://www.mit.edu/~yandros/doc/specs/fcgi-spec.html#S5.4 I
came across this:

> A Web server aborts a FastCGI request when an HTTP client closes its 
> transport connection while the FastCGI request is running on behalf of that 
> client. 

> The Web server sends a FCGI_ABORT_REQUEST record to abort a request. After 
> receiving {FCGI_ABORT_REQUEST, R}, the application responds as soon as 
> possible with {FCGI_END_REQUEST, R, {FCGI_REQUEST_COMPLETE, appStatus}}. This 
> is truly a response from the application, not a low-level acknowledgement 
> from the FastCGI library.

As far as I can see the patch breaks out of the main while after sending the
FCGI ABORT, not waiting for the FCGI_END_REQUEST response. I tried to change it
a bit to allow this flow, but from my test I never get FCGI_END_REQUEST because
mod_proxy_fcgi stops before it finding out that the client dropped the
connection (probably when it tries to flush the output).

I also didn't manage to see the FCGI backend stopping after the FCGI ABORT is
sent, but maybe I am missing something trivial. I'll try to work on this code
more during the next days.

Jason: just want to make it clear, your patch looks good, I just want to make
sure that we don't miss any corner case :)

Thanks!

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to