Re: [jetty-dev] SPDY window size update, a bug?

[Simone Bordet <sbordet@xxxxxxxxxxx>]

Hi,

On Wed, Mar 28, 2012 at 07:40, Will <wglozer@xxxxxxxxx> wrote:
> Hello,
>
> I have been playing around with Jetty's shiny new SPDY support and ran
> into a problem where my SPDY sessions were timing out, apparently
> because a PING response wasn't being sent to the client (Chrome 17).
>
> I tracked this down to StandardSession, where a flush call was being
> skipped with two items in the queue, a DataFrame and the PingFrame
> response. The flush was skipped because DataFrameBytes.getByteBuffer()
> was returning null, which in turn was caused by the Stream window size
> being <= 0.
>
> When the complete() method of DataFrameBytes is called it calls
> Stream.updateWindowSize(-length):
>
> https://github.com/eclipse/jetty.project/blob/jetty-8/jetty-spdy/spdy-core/src/main/java/org/eclipse/jetty/spdy/StandardSession.java#L1037
>
> Is that correct? I don't follow the logic for updating the window size
> that way, but I am very new to both SPDY and the Jetty code base.
> Removing the window size update does fix the client PING timeouts.

I am not sure that flow control is implemented in Chromium/Chrome 17,
so perhaps we need a switch to turn it off on server side too.
I think the update of the window size is right, but we're stalling
PING frames (and other channels too), so that looks like a bug.
Investigating.

Simon
-- 
http://cometd.org
http://intalio.com
http://bordet.blogspot.com
----
Finally, no matter how good the architecture and design are,
to deliver bug-free software with optimal performance and reliability,
the implementation technique must be flawless.   Victoria Livschitz