Another update, I've pushed what I have to a branch called `issue/CAMEL-12871` the implementation of the stub server and integration test for two cases. In both cases streaming seems to have continued, that is, I haven't reproduced the issue. I also included a fix to release listeners/subscribers and close channels when stopping the SubscriptionHelper. I think this might help also.
My hypothesis is that a way to reproduce the issue is to half-close the TCP connection, i.e. without getting a packet with the FIN flag from the server, which might be the case in the reproducer provided by Andy. But I don't think I have that capability using Java. Next, I think I'll try to reproduce this using Andy's steps, i.e. using the firewall to simulate an outage and dig into the network traffic to determine what actually happens. I'd like to get feedback, i.e. am I simulating the outage correctly via the stub server, and do the tests prove correct behaviour. zoran On Sat, Dec 12, 2020 at 7:22 AM Jean-Baptiste Onofre <j...@nanthrax.net> wrote: > > Thanks for the update Zoran. > > Let me take a look on your gist. > > Regards > JB > > > Le 10 déc. 2020 à 13:46, Zoran Regvart <zo...@regvart.com> a écrit : > > > > Hi JB, > > split off into a separate thread, yeah, thanks, I could use some help/ideas. > > > > I'm trying to create an integration test, using WireMock and > > WiremockNetworkTrafficListener[1] to simulate the issue reported on > > the users mailing list[2]. > > > > Doesn't seem like I will be able to, so I might try another approach. > > The crux of the issue seems to be that WireMock is not really well > > suited to simulate CometD (hard to simulate pushes) and that just > > closing the socket via WiremockNetworkTrafficListener doesn't lead to > > the same error scenario, i.e. the HTTP client just retries (from what > > it seems). > > > > It's hard for me to work without a test, and fiddling with the > > firewall to block the connection takes too much time in each iteration > > and is hard to do reliably. > > > > I'll try to work something out, this[3] is what I have right now, not > > really something to push and I am changing it trying to find the right > > set of circumstances to trigger the error case. > > > > zoran > > > > [1] > > https://github.com/tomakehurst/wiremock/blob/d43540a4a4421c49a2b866db5a19dca5d428f85b/src/main/java/com/github/tomakehurst/wiremock/http/trafficlistener/WiremockNetworkTrafficListener.java > > [2] > > https://lists.apache.org/thread.html/rb54e7c161eef79ed1142caaf549a0ab568697ec82f246c7fe9132ee7%40%3Cusers.camel.apache.org%3E > > [3] https://gist.github.com/zregvart/f7697cfd3c325afa6f941dfd32087bd0 > > > > > > On Thu, Dec 10, 2020 at 12:51 PM JB Onofré <j...@nanthrax.net> wrote: > >> > >> Hey Zoran > >> > >> Do you need help on the sales force streaming issue ? > >> > >> Regards > >> JB > >> > >>> Le 10 déc. 2020 à 12:50, Zoran Regvart <zo...@regvart.com> a écrit : > >>> > >>> I was hoping to fix the Salesforce streaming issue reported on the > >>> users mailing list, I guess that'll have to wait for 3.7.1 as I'm > >>> having issues reproducing this. > >>> > >>> zoran > >>> > >>>> On Thu, Dec 10, 2020 at 12:10 PM Andrea Cosentino <anco...@gmail.com> > >>>> wrote: > >>>> > >>>> I just fixed some documentation issues with google stream components. > >>>> > >>>> Green light for me. > >>>> > >>>> Il giorno gio 10 dic 2020 alle ore 07:56 Claus Ibsen > >>>> <claus.ib...@gmail.com> > >>>> ha scritto: > >>>> > >>>>> Hi Gregore > >>>>> > >>>>> The FTP tests are fixed. > >>>>> We are ready for cutting the release. > >>>>> > >>>>> On Thu, Dec 3, 2020 at 9:37 PM Gregor Zurowski > >>>>> <gre...@list.zurowski.org> > >>>>> wrote: > >>>>>> > >>>>>> Yes, I am available for cutting the release as soon as things are > >>>>>> ready. > >>>>>> > >>>>>> Thanks, > >>>>>> Gregor > >>>>>> > >>>>>> > >>>>>> On Thu, Dec 3, 2020, 11:03 AM Claus Ibsen <claus.ib...@gmail.com> > >>>>>> wrote: > >>>>>> > >>>>>>> Hi > >>>>>>> > >>>>>>> I am still working on the csimple language. There are some more work > >>>>>>> needed. > >>>>>>> And the JIRA has 33 tickets for 3.7. It would be good to get some more > >>>>>>> work done, or move to 3.8. > >>>>>>> I am afraid that we need another week of development to be ready. > >>>>>>> > >>>>>>> Gregor if you are listening when is a good time for you to cut the > >>>>> release > >>>>>>> ? > >>>>>>> > >>>>>>> > >>>>>>> On Mon, Nov 23, 2020 at 1:15 PM Claus Ibsen <claus.ib...@gmail.com> > >>>>> wrote: > >>>>>>>> > >>>>>>>> Hi > >>>>>>>> > >>>>>>>> Next month (december) we should aim to get a new 3.7 LTS release out. > >>>>>>>> > >>>>>>>> So lets start to close down on remaining work, and get CI stable. > >>>>>>>> > >>>>>>>> I am currently working on compiled simple language (preview) for 3.7 > >>>>>>>> which is an effort that takes all of this week. > >>>>>>>> > >>>>>>>> The JIRA has 43 tickets assigned > >>>>>>>> > >>>>>>> > >>>>> https://issues.apache.org/jira/issues/?jql=project%20%3D%20CAMEL%20AND%20fixVersion%20%3D%203.7.0%20AND%20statusCategory%20%3D%20new > >>>>>>>> > >>>>>>>> Some of those can likely be moved to 3.8, as usual we dont have time > >>>>>>>> to complete all those tickets. > >>>>>>>> > >>>>>>>> Lets try to take 2 weeks of development and then see if we are ready > >>>>>>>> for cutting the RC. > >>>>>>>> > >>>>>>>> > >>>>>>>> -- > >>>>>>>> Claus Ibsen > >>>>>>>> ----------------- > >>>>>>>> http://davsclaus.com @davsclaus > >>>>>>>> Camel in Action 2: https://www.manning.com/ibsen2 > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> -- > >>>>>>> Claus Ibsen > >>>>>>> ----------------- > >>>>>>> http://davsclaus.com @davsclaus > >>>>>>> Camel in Action 2: https://www.manning.com/ibsen2 > >>>>>>> > >>>>> > >>>>> > >>>>> > >>>>> -- > >>>>> Claus Ibsen > >>>>> ----------------- > >>>>> http://davsclaus.com @davsclaus > >>>>> Camel in Action 2: https://www.manning.com/ibsen2 > >>>>> > >>> > >>> > >>> > >>> -- > >>> Zoran Regvart > >> > > > > > > -- > > Zoran Regvart > -- Zoran Regvart
