I believe, that people in a specialized Tomcat forum may be able to help you a lot better.
As a last hint, if all else fails, you can still experiment with a timeout on the client. See this thread: http://groups.google.com/group/google-web-toolkit/browse_thread/thread/11960c28e1a2d9eb?pli=1 Or set something like a content aware load balancer in front, to balance between your Apache and Tomcat without acting like a proxy, if that's possible at all, and if it works with your special PHP setup. Sorry I couldn't help you with more specifics, but it's been some years since I last used a Tomcat+Apache setup... Chris On Feb 22, 7:19 pm, Fran <fra...@gmail.com> wrote: > I think that I will need a iframe that loads tomcat page. But this > solution isnt good > > On 21 feb, 16:49, Fran <fra...@gmail.com> wrote: > > > > > it dont work, apache cant conect totomcatwith this changes. The > > login msn is loading indefinitely... And error logs dont show nothing, > > excepttomcatthat says: > > > org.apache.tomcat.util.net.NioSelectorPool getSharedSelector > > INFO: Using a shared selector for servlet write/read > > > Here the changes... > > Added this lines in event method: > > > if (event.getEventType() == CometEvent.EventType.BEGIN) { > > event.setTimeout(30*1000); > > } > > > if (event.getEventSubType() == CometEvent.EventSubType.TIMEOUT) { > > writeResponse(event.getHttpServletResponse(), > > "timeout"); > > event.close(); > > } > > > server.xml (tomcat): > > > <Connector > > connectionTimeout="20000" > > port="8081" > > protocol="org.apache.coyote.http11.Http11NioProtocol" > > maxThreads="5" > > acceptorThreadCount="2" > > redirectPort="8443" > > socket.directBuffer="false" /> > > > httpd.conf (apache): > > > keepalive Off > > ProxyRequests Off > > SetEnv force-proxy-request-1.0 1 > > SetEnv proxy-nokeepalive 1 > > > <Proxy *> > > Order deny,allow > > Allow from all > > </Proxy> > > > ProxyPass > > /msn/com.msn/messengerhttp://localhost:8081/msn/com.msn/messenger > > ProxyPassReverse > > /msn/com.msn/messengerhttp://localhost:8081/msn/com.msn/messenger > > > On 21 feb, 16:29, Chris Lercher <cl_for_mail...@gmx.net> wrote: > > > > I expect that your application works, if you usetomcatonly, because > > > many browsers will wait for the response for a very long time. In that > > > case, you don't need a timeout on your server. > > > > But when there's any proxy between your client and the server (and > > > that's also the case when the client connects to the Internet via a > > > proxy), then it often fails. Also expected behaviour. > > > > I'd take a look at this:http://tomcat.apache.org/tomcat-6.0-doc/aio.html > > > > It says you can set a timeout like this: > > > event.setTimeout(30*1000) > > > > I'd do this for EventType.BEGIN. > > > > Then you can test for EventType.TIMEOUT (the documentation says, that > > > you may need the > > > org.apache.catalina.valves.CometConnectionManagerValve for this). > > > > On timeout, send some dummy response. The client should recognize the > > > response as a timeout message, and retry. > > > > So, in effect I imagine it could look something like this (I can't try > > > it here, so there may be some mistakes): > > > > public void event(CometEvent event) throws IOException, > > > ServletException { > > > if (event.getEventType() == CometEvent.EventType.BEGIN) { > > > event.setTimeout(30*1000); > > > } > > > if (event.getEventType() == CometEvent.EventType.READ) { ...} > > > if (event.getEventType() == CometEvent.EventType.TIMEOUT) { > > > writeResponse(response, "timeout"); > > > event.close(); > > > } > > > > } > > > > On Feb 21, 3:54 pm, Fran <fra...@gmail.com> wrote: > > > > > Chris, the big problem is that the aplication works fully intomcat > > > > but I use apache to run it with the servlet intomcat, it cant work. > > > > I only need to emulate that the execution in apache is the same that > > > > intomcat > > > > > On 21 feb, 15:50, Fran <fra...@gmail.com> wrote: > > > > > > I based the code in a GWT book. > > > > > Please take a look > > > > > athttp://217.13.89.62/messenger%20nonblocking%20calls.pdf > > > > > next the phrase at the first page: "Using Server-Side Advanced IO" > > > > > > Thanks for your time > > > > > > On 21 feb, 15:38, Chris Lercher <cl_for_mail...@gmx.net> wrote: > > > > > > > Using writer.flush() is absolutely ok, it's just not enough. Taking > > > > > > a > > > > > > quick look at the code you posted, it looks ok, because after > > > > > > calling > > > > > > writer.flush(), it always calls event.close(). So, no problem here. > > > > > > But: I don't see any timeout in the code. The server must also call > > > > > > event.close() after some timeout - which must be shorter than your > > > > > > proxy's and your browser's timeout. (Some people say, it should in > > > > > > any > > > > > > case be less than a minute.) > > > > > > > So either you can configure the timeout somewhere intomcat(please > > > > > > refer to thetomcatdocumentation), or you'll have to send something > > > > > > back manually after some time. > > > > > > When the client receives such a dummy response, it must open a new > > > > > > request (that's the long polling principle). > > > > > > > On Feb 21, 3:28 pm, Fran <fra...@gmail.com> wrote: > > > > > > > > Yes Chris, I use writer.flush() > > > > > > > I cant use it at mod_proxy? What can I do? > > > > > > > > Thanks > > > > > > > > On 21 feb, 14:39, Chris Lercher <cl_for_mail...@gmx.net> wrote: > > > > > > > > > Here are some things you can check - maybe one of these helps > > > > > > > > (?) > > > > > > > > > - In your servlet: Do you expect that anything gets flushed > > > > > > > > (reliably), before you call the close() method on CometEvent? > > > > > > > > If yes, > > > > > > > > then what you're actually trying to do is _streaming_ (which > > > > > > > > doesn't > > > > > > > > work). Calling writer.flush() is certainly not enough, since > > > > > > > > mod_proxy > > > > > > > > has no way to know that writer.flush() has been called. > > > > > > > > > - Did you try to set "no-cache" headers etc.? > > > > > > > > > On Feb 21, 1:11 pm, Fran <fra...@gmail.com> wrote: > > > > > > > > > > I tested the mod_proxy_http, in the last post you can see the > > > > > > > > > config > > > > > > > > > that I used. > > > > > > > > > But in Comet HTTP streaming, I found mod_proxy blocked the > > > > > > > > > response > > > > > > > > > until server close the > > > > > > > > > stream (after timeout), then send the whole response to > > > > > > > > > client at > > > > > > > > > once. > > > > > > > > > > In the first time I was wrong because I thought that if I put > > > > > > > > > timeout=1 I found the solution but Its not real because the > > > > > > > > > server > > > > > > > > > close the conection after timeout > > > > > > > > > > In the five post of mine is an example of the problem. > > > > > > > > > > This dude has a similar > > > > > > > > > problem:http://forums.java.net/jive/thread.jspa?threadID=41377 > > > > > > > > > > I am hopeless :( > > > > > > > > > > On 21 feb, 03:21, dablack <david.blackwell...@gmail.com> > > > > > > > > > wrote: > > > > > > > > > > > Sorry to lead you on a wild goose chase. I guess you will > > > > > > > > > > have to use > > > > > > > > > > the Apache mod_proxy_http module with theTomcatNIO > > > > > > > > > > connector using > > > > > > > > > > the http protocol after all. I don't have any experience > > > > > > > > > > with the > > > > > > > > > > mod_proxy_http module at all so not really anything I can > > > > > > > > > > help you > > > > > > > > > > with. > > > > > > > > > > > Good luck. > > > > > > > > > > > On Feb 20, 7:21 pm, Fran <fra...@gmail.com> wrote: > > > > > > > > > > > > The problem persist with the news changes. > > > > > > > > > > > Seems that apache cant connect totomcat:( > > > > > > > > > > > The error log of apache and mod_jk dont show nothing. Not > > > > > > > > > > > ever nio conection > > > > > > > > > > > stablished > > > > > > > > > > > > 2010/2/21 dablack <david.blackwell...@gmail.com> > > > > > > > > > > > > > I borrowed that connector configuration from a website > > > > > > > > > > > > I came across. > > > > > > > > > > > > I'm not sure why port=0 either. If I put more research > > > > > > > > > > > > into the NIO > > > > > > > > > > > > connector I could probably find out. Why don't you try > > > > > > > > > > > > adding the > > > > > > > > > > > > lines, > > > > > > > > > > > > > enableLookups="false" > > > > > > > > > > > > scheme="http" > > > > > > > > > > > > > to the NIO connector configuration in the server.xml > > > > > > > > > > > > file that I gave > > > > > > > > > > > > you earlier. > > > > > > > > > > > > > As far as accessing the ajp service > > > > > > > > > > > > usinghttp://localhost:8009/servlet, > > > > > > > > > > > > that will not work. Your browser uses the http protocol > > > > > > > > > > > > and the 8009 > > > > > > > > > > > > port is being serviced by the ajp protocol. You should > > > > > > > > > > > > still have your > > > > > > > > > > > > connector using the http protocol configured to listen > > > > > > > > > > > > on port 8081. > > > > > > > > > > > > > Hopefully the scheme line above will help the NIO > > > > > > > > > > > > connector > > > > > > > > > > > > communicate back to Apache. > > > > > > > > > > > > > On Feb 20, 4:45 pm, Fran <fra...@gmail.com> wrote: > > > > > > > > > > > > > I test your config, dablack, but I cant connect to > > > > > > > > > > > > > the servlet. > > > > > > > > > > > > > With this config I cant access not even > > > > > > > > > > > > > to:http://localhost:8009/servlet > > > > > > > > > > > > > that istomcatdirectly without apache > > > > > > > > > > > > > > On 20 feb, 23:36, Fran <fra...@gmail.com> wrote: > > > > > > > > > > > > > > > I cant undestand why this config: > > > > > > > > > > > > > > > <Connector protocol="AJP/1.3" port="0" > > > > > > > > > > > > > > channelNioSocket.port="8009" > > > > > > > > > > > > > > channelNioSocket.maxThreads=" > > > > > > > > > > > > > > 150" > > > > > > > > > > > > > > channelNioSocket.maxSpareThreads="50" > > > > > > > > > > > > > > channelNioSocket.minSpareThreads="25" > > > > > > > > > > > > > > channelNioSocket.bufferSize="16384" /> > > > > > > > > > > > > > > > Why port 0 ? > > > > > > > > > > > > > > > 2010/2/20 dablack <david.blackwell...@gmail.com> > > > > > > > > > > > > > > > > Fran, > > > > > > > > > > > > > > > > I'm really not an expert on setting up the > > > > > > > > > > > > > > > connection between Apache > > > > > > > > > > > > > > > andTomcatbut I believe that the connections only > > > > > > > > > > > > > > > operate in two > > > > > > > > > > > > > > > protocols: http or ajp. NIO is aTomcatconnector > > > > > > > > > > > > > > > that can > > > > > > > > > > > > communicate > > > > > > > > > > > > > > > in either http or ajp. Because of that, I don't > > > > > > > > > > > > > > > see any reason that > > > > > > > > > > > > > > > the Apache mod_jk connector module couldn't > > > > > > > > > > > > > > > communicate with the NIO > > > > > > > > > > > > > > > connector. Since I don't use an NIO connector I > > > > > > > > > > > > > > > could be wrong, but > > > > > > > > > > > > > > > that is the way I see it. As a starting point, > > > > > > > > > > > > > > > try these > > > > > > > > > > > > > > > configurations: > > > > > > > > > > > > > > > > server.xml > > > > > > > > > > > > > > > > <Connector protocol="AJP/1.3" port="0" > > ... > > read more » -- You received this message because you are subscribed to the Google Groups "Google Web Toolkit" group. To post to this group, send email to google-web-tool...@googlegroups.com. To unsubscribe from this group, send email to google-web-toolkit+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-web-toolkit?hl=en.
