On Thu, Nov 15, 2018 at 8:45 AM Oleg Kalnichevski <[email protected]> wrote:
> On Sun, 2018-11-11 at 18:30 -0700, Gary Gregory wrote: > > Hi All: > > > > ... > > > I am looking for guidance here. I suppose this could be a bug in the > > example, or in the core, or that the example could/should be adjusted > > to > > avoid this unfortunate sequence. Again, the IO rector parameters I > > used are > > only to demonstrate the problem. In my set up, I use the default > > select > > timeout of 1 second and reactor IO thread counts of 2. > > > > Thank you, > > Gary > > > Oh, it was a nasty one. It is just sheer luck that the problem was much > easier to reproduce with non-persistent connections and out of laziness > I opted for AB (which uses non-persistent HTTP/1.0 by default) instead > of JMeter. > > I believe the culprit was this race condition: > > > https://github.com/apache/httpcomponents-core/commit/f5590e471f46892787df01fca2a6e8faf26fd454 > > I also fixed incorrect handling of incoming HTTP/1.0 requests by the > reverse proxy example. > > Please test the latest snapshot with your application and let me know the > outcome. > Thank you Oleg, I am testing... My initial app test failed (I made the same change to my app as you made to the sample to use a distinct HttpContext for outgoing message exchanges) so I will review and try to reproduce with the example... Gary > > Cheers > > Oleg > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
