Hi Rémy, Many thanks for your tip!
I added this parameter to the Connector and now it works! So it seems that the new async IO was causing the issue. First feeling is that response times are a little bit lower now but still better than HTTP/1.1 Should we see this as an issue in our code which requires (for some reason) that scripts are loaded in some sequence or is this something that needs to be fixed? Tnx again for helping us to resolve this, much appreciated! Regards, /Tom On Thu, May 23, 2019 at 10:04 PM Rémy Maucherat <r...@apache.org> wrote: > On Thu, May 23, 2019 at 8:44 PM Tom Coudyzer <tcdm...@gmail.com> wrote: > > > Hi, > > > > I tested with the 9.0.21-dev version and the issue is still there. > > I attached two screenshots on what is happening in Chrome (same behavior > > is present in Firefox). In Internet Explorer 11 there are no errors. > > > > While I wait on some feedback will try to find out how I can enable the > > logging (with more info). As I told earlier Tomcat is embedded in an > > application and it looks something has changed to the standard logging > > mechanism. Since I was only asked recently to have a look at this > product I > > need to find this out.Think we'll need this so that we can see what's > going > > on. > > > > When I need to try something or you have some ideas, just let me know. > > Will try them out. > > > > Ok, so the new async IO might be the cause of your problem, you should set > useAsyncIO="false" on the connector. > > I don't know what your page is doing, but I pulled the HTTP/2 test page we > did demo with at conferences and if I insist I can see problems. > This is caused by the check line 1530 in Http2UpgradeHandler, the state > being CLOSED_RX (so no window updating normally). If I comment it out, I > see no error or problems anymore in the logs, and things work reliably. So > maybe it's a processing ordering problem caused by async. I added some sync > in the parser and it works too, not sure why right now. > > Rémy > > > > > > Thanks already! > > > > /Tom > > > > > > On Thu, May 23, 2019 at 3:42 PM Mark Thomas <ma...@apache.org> wrote: > > > >> On 23/05/2019 11:52, Tom Coudyzer wrote: > >> > Hi Mark, > >> > > >> > Thanks for the info. > >> > > >> > If you say where we can download the test build we are happy to do > that > >> and > >> > see if it makes any difference. > >> > >> http://people.apache.org/~markt/dev/v9.0.21-dev-d6d3b31/ > >> > >> > No problem if any (serious) issues can occur, we'll have the fire > >> > extinguishers ready :-) > >> > >> Great. I just wanted you to be aware that this only a test build and is > >> in no way an official ASF release. > >> > >> Let us know how you get on. > >> > >> Thanks, > >> > >> Mark > >> > >> > >> > > >> > Once again many thanks for your help, much appreciated! > >> > > >> > Regards, > >> > Tom > >> > > >> > On Thu, May 23, 2019 at 11:09 AM Mark Thomas <ma...@apache.org> > wrote: > >> > > >> >> On 23/05/2019 09:04, Tom Coudyzer wrote: > >> >>> Hi, > >> >>> > >> >>> We analyzed the problem further and we have set up another version > >> which > >> >> is > >> >>> using Tomcat 9.0.19. We don't have the issue anymore. So it looks > like > >> >> it's > >> >>> caused by something that has changed between the two versions. Based > >> on > >> >> the > >> >>> info I have it looks like Chrome is stalling and getting too many > >> >> responses > >> >>> at the same time. In the meantime we activated already the access > log > >> and > >> >>> all files in that file are sent with status 200. So it looks like a > >> >> client > >> >>> / browser side issue. > >> >>> > >> >>> Does this information give more information on what could be causing > >> the > >> >>> problem or which parameter we should set up / change or do you still > >> need > >> >>> more information? > >> >> > >> >> Given the issue appears between 9.0.19 and 9.0.20 a Tomcat issue > looks > >> >> more likely than a browser issue. > >> >> > >> >> If I point you towards a test build (no guarantees, use at your own > >> >> risk, if your server catches fire don't blame me etc.) for 9.0.21 > could > >> >> you test that as there are some HTTP/2 changes in 9.0.20 that made a > >> >> long standing bug more likely that have been fixed in 9.0.21-dev. > >> >> > >> >> If you still see the issue then we can look at specific debug logging > >> >> and/or isolated test cases. > >> >> > >> >> Mark > >> >> > >> >> > >> >>> > >> >>> On Wed, May 22, 2019 at 5:01 PM Mark Thomas <ma...@apache.org> > wrote: > >> >>> > >> >>>> On 22/05/2019 15:47, Tom Coudyzer wrote: > >> >>>>> Hi, > >> >>>>> > >> >>>>> We wanted to upgrade our application to start using HTTP/2. We > added > >> >> the > >> >>>>> necessary and we see that the browser is using HTTP/2 in the > >> browsers' > >> >>>>> development tools. > >> >>>>> > >> >>>>> However since we activated it we get random CONNECTION_CLOSED 200 > >> >> Errors > >> >>>> in > >> >>>>> Chrome. It's not always on the same files and sometimes there are > >> more, > >> >>>>> sometimes there are less, sometimes it works. When we disable > HTTP/2 > >> >>>>> (remove the upgradeprotocol tag) everything works fine. > >> >>>>> > >> >>>>> We are running Tomcat 9.0.20 (x64) on a Windows Server 2008 R2 > >> server. > >> >>>>> > >> >>>>> Did we configure something incorrectly, is this an HTTP/2 issue or > >> >> should > >> >>>>> we look at network issues on our end? > >> >>>>> > >> >>>>> If you need more information or we need to run something to be > able > >> to > >> >>>>> troubleshoot it better please let me know. > >> >>>>> > >> >>>>> Help is much appreciated ! > >> >>>> > >> >>>> You can try enabling debug logging if the issue is fairly easy to > >> >>>> reproduce. That might shed some light on what Tomcat is doing and > >> why. > >> >>>> > >> >>>> There are also some HTTP/2 fixes due in the next set of releases > that > >> >>>> might help. > >> >>>> > >> >>>> 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 > >> > >> > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > > For additional commands, e-mail: users-h...@tomcat.apache.org >
