Stefan,

You said this was an 8.x to 9.x upgrade, so if the F5 was existing, it
shouldn't be an issue unless there are infrastructure changes at the same
time for upgrade.  We use F5 and the only time we have seen something
similar was where a perimeter f/w was dropping an established connection
(due to inactivity) within the keep alive time.   When the f/w does that
the browser does not know the connection it has is stuffed.

tcpdump is the best approach.




On 28 January 2014 08:45, Stefan Magnus Landrø <[email protected]>wrote:

> Thanks, Simone! I'm afraid I have to agree with you - this probably won't
> fix my issue. I'll do some tcpdumping in the next few days - I'll keep you
> posted. This might as well turn out being a f5 big ip issue.
>
> Stefan
>
>
> 2014-01-27 Simone Bordet <[email protected]>
>
> Hi,
>>
>> On Fri, Jan 24, 2014 at 9:36 PM, "" <[email protected]> wrote:
>> > Good find and a _critical_ bug in my view.
>> >
>> > A change that came with 9.0 as it was this in 8.x :-
>> >
>> >                      _header.put(CONNECTION_KEEP_ALIVE);
>> >                      if (connection!=null)
>> >                      {
>> >                          _header.setPutIndex(_header.putIndex()-2);
>> >                          _header.put((byte)',');
>> >                          _header.put(connection.toString().getBytes());
>> >                          _header.put(CRLF);
>> >                      }
>> >
>> > which had less logic, shorter lines... Yes, it copied 2 more bytes than
>> it
>> > needed to but I would prefer less logic/shorter code over saving 2
>> bytes.
>> >
>> > To fix this, you could _try_ turning off one connect on the f5 if you
>> don't
>> > need it for other reasons.
>> >
>> > We are also behind an F5, and are just about to start looking at moving
>> from
>> > 8.x to 9.x.   ... so thanks ... will raise on support.
>>
>> I filed https://bugs.eclipse.org/bugs/show_bug.cgi?id=426739 and fixed
>> it.
>>
>> Note that this is a corner case that should happen very rarely, only
>> when the "Connection" header gets some custom value that is not
>> "close" or "upgrade" or "keep-alive".
>>
>> Therefore, I am dubious that the original problem (curl hanging for
>> chunked responses) is fixed by this fix.
>>
>> Let us know if that is the case, though.
>>
>> --
>> Simone Bordet
>> ----
>> http://cometd.org
>> http://webtide.com
>> http://intalio.com
>> Developer advice, training, services and support
>> from the Jetty & CometD experts.
>> Intalio, the modern way to build business applications.
>> _______________________________________________
>> jetty-users mailing list
>> [email protected]
>> https://dev.eclipse.org/mailman/listinfo/jetty-users
>>
>
>
>
> --
> BEKK Open
> http://open.bekk.no
>
> TesTcl - a unit test framework for iRules
> http://testcl.com
>
> _______________________________________________
> jetty-users mailing list
> [email protected]
> https://dev.eclipse.org/mailman/listinfo/jetty-users
>
>
_______________________________________________
jetty-users mailing list
[email protected]
https://dev.eclipse.org/mailman/listinfo/jetty-users

Reply via email to