Hi Andre, I have not run all my tests yet but it looks like you were correct. The Amazon load balancer appears to have been the culprit. I will let you know if it turns out otherwise.
Thank you for pushing me to take the load balancer out. Cheers! Szymon On 28 November 2016 at 13:45, Szymon Czaja <szymonpiotrcz...@gmail.com> wrote: > Completely agree. I am in the process of setting up a direct connection to > Tomcat bypassing load balancer. Will give an update soon. > > Szymon > > On 28 November 2016 at 13:40, André Warnier (tomcat) <a...@ice-sa.com> > wrote: > >> On 28.11.2016 14:10, Szymon Czaja wrote: >> >>> I will run a test in the meantime to check but this is very much >>> unlikely, >>> I can see on the Amazon console that the load balancer CPU is hardly >>> doing >>> anything with peaks at 2%. Let's assume load balancer is not an issue. >>> >> >> I am not saying that the proxy cannot handle it. But maybe it is limiting >> the number of connections that it forwards, if they are all coming from the >> same client IP ? >> (Avoid DoS attacks, that kind of thing..) >> Still a guess. But netstat on both sides may be telling you more. >> >> If it is tomcat that could not handle the load, then you'd probably still >> see the same number of connections on the tomcat server side, at the TCP >> level. >> That tomcat would be able to "service" these connections is another >> matter, but the connections themselves should be there. >> >> >> >>> Szymon >>> >>> On 28 November 2016 at 12:29, André Warnier (tomcat) <a...@ice-sa.com> >>> wrote: >>> >>> On 28.11.2016 12:52, Szymon Czaja wrote: >>>> >>>> Hi, >>>>> I have updated my question on SO. I have noticed that the number of >>>>> ESTABLISHED connections goes up on the client after few minutes. I >>>>> expected >>>>> the same on the server which does not seem to be the case. Any ideas? >>>>> >>>>> >>>> Just a guess without looking very deep into your data : you say >>>> somewhere >>>> that the requests go through a proxy. Maybe the client "established" TCP >>>> connections are with that proxy, while the ones you see on the tomcat >>>> server are the connections from the proxy ? >>>> (In other words, it is the proxy that is the bottleneck ?) >>>> >>>> >>>> >>>> >>>> Szymon >>>>> >>>>> On 28 November 2016 at 10:05, Mark Thomas <ma...@apache.org> wrote: >>>>> >>>>> On 28/11/2016 09:53, Szymon Czaja wrote: >>>>> >>>>>> >>>>>> Hi, >>>>>>> I have asked the question on StackOverflow but I am not getting much >>>>>>> response: >>>>>>> >>>>>>> http://stackoverflow.com/questions/40793335/why-tomcat- >>>>>>> does-not-pass-a-tsung-performance-test-at-40-requests-per-second >>>>>>> >>>>>>> Could anyone help me understand why is Tomcat unable to keep up with >>>>>>> the >>>>>>> processing when running Tsung yet Apache Bench tests do not relveal >>>>>>> any >>>>>>> scalability issues? >>>>>>> >>>>>>> >>>>>> I'd recommend taking some thread dumps and looking at netstat output >>>>>> to >>>>>> try and figure out what is going on. >>>>>> >>>>>> Mark >>>>>> >>>>>> >>>>>> --------------------------------------------------------------------- >>>>>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >>>>>> For additional commands, e-mail: users-h...@tomcat.apache.org >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >>>> For additional commands, e-mail: users-h...@tomcat.apache.org >>>> >>>> >>>> >>> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org >> For additional commands, e-mail: users-h...@tomcat.apache.org >> >> >
