Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: UK time here. Still trying to get Ethereal to trace local (intra machine) traffic. I'm not around tomorrow so it may be Monday before there's anything to dissect. Monday came and went. Is anyone else interested in the detail of what follows (analysis of packet dumps) or should we take it off list? Anyway I've just now had time to look at the packet dump I captured on Thursday. The sequence appears to be Domino: Sends first request to Tomcat Tomcat: Replies correctly Domino: Sends second request to Tomcat, gets nothing back (there's no data flow from Tomcat at all at this stage) Domino: Periodically retries sending the request packet That's as far as I've got and I'm just wondering whether it'd be nice to write an Ethereal packet dissector for AJP 1.3 or whether that would count as displacement activity. Is there anything specific I should be looking for Henri? -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Andy Armstrong wrote: UK time here. Still trying to get Ethereal to trace local (intra machine) traffic. I'm not around tomorrow so it may be Monday before there's anything to dissect. Monday came and went. Is anyone else interested in the detail of what follows (analysis of packet dumps) or should we take it off list? Anyway I've just now had time to look at the packet dump I captured on Thursday. The sequence appears to be Domino: Sends first request to Tomcat Tomcat: Replies correctly Domino: Sends second request to Tomcat, gets nothing back (there's no data flow from Tomcat at all at this stage) Domino: Periodically retries sending the request packet That's as far as I've got and I'm just wondering whether it'd be nice to write an Ethereal packet dissector for AJP 1.3 or whether that would count as displacement activity. Is there anything specific I should be looking for Henri? There is allready an ajp13 disector for ethereal ;) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: That's as far as I've got and I'm just wondering whether it'd be nice to write an Ethereal packet dissector for AJP 1.3 or whether that would count as displacement activity. Is there anything specific I should be looking for Henri? There is allready an ajp13 disector for ethereal ;) OK, I'll stop trying to get my head round how to write one then :) -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Henri Gomez wrote: That's as far as I've got and I'm just wondering whether it'd be nice to write an Ethereal packet dissector for AJP 1.3 or whether that would count as displacement activity. Is there anything specific I should be looking for Henri? There is allready an ajp13 disector for ethereal ;) OK, I'll stop trying to get my head round how to write one then :) The ajp13 disector was introduced in ethereal some months ago - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: The ajp13 disector was introduced in ethereal some months ago And I now have a copy of it rather than last year's model :) Anyway, the conversation looks like this in summary and a (naive) analysis of the contents of the packets doesn't yield anything obvious. - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST Should there be anything else before the second FORWARD REQUEST? The Domino end seems to spin in first while loop inside ajp_send_request() because ajp_connection_tcp_send_message() keeps returning JK_TRUE without ae-sd becoming = 0. Debug indicates that ae-sd retains its initial value on successive trips round the loop. -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Henri Gomez wrote: The ajp13 disector was introduced in ethereal some months ago And I now have a copy of it rather than last year's model :) Anyway, the conversation looks like this in summary and a (naive) analysis of the contents of the packets doesn't yield anything obvious. - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST - FORWARD REQUEST Should there be anything else before the second FORWARD REQUEST? The Domino end seems to spin in first while loop inside ajp_send_request() because ajp_connection_tcp_send_message() keeps returning JK_TRUE without ae-sd becoming = 0. Debug indicates that ae-sd retains its initial value on successive trips round the loop. I've got : - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE (REUSEP:1) May be something to do with chunk encoding which has been modified in ajp13/jk (after your works on domino I think) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE (REUSEP:1) OK, that's enough of a difference to explain what I'm seeing... May be something to do with chunk encoding which has been modified in ajp13/jk (after your works on domino I think) Well so far I've taken the old Domino Connector source and built it against the latest version of ajp13/jk (as of last week). Should that not 'just work'? -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Henri Gomez wrote: - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE (REUSEP:1) OK, that's enough of a difference to explain what I'm seeing... May be something to do with chunk encoding which has been modified in ajp13/jk (after your works on domino I think) Well so far I've taken the old Domino Connector source and built it against the latest version of ajp13/jk (as of last week). Should that not 'just work'? Nope, there was works on the chunked and header len in ajp13 java side which are probably not in old domino code... Take a look at what is on IIS which should be at right level - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Henri Gomez wrote: - FORWARD REQUEST - SEND HEADERS - SEND BODY CHUNK - END RESPONSE (REUSEP:1) OK, that's enough of a difference to explain what I'm seeing... May be something to do with chunk encoding which has been modified in ajp13/jk (after your works on domino I think) Well so far I've taken the old Domino Connector source and built it against the latest version of ajp13/jk (as of last week). Should that not 'just work'? BTW, we're now focusing on jk2, so what about working on jk2 for domino. The requirement will be having apr for your box (windows ?) - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: Nope, there was works on the chunked and header len in ajp13 java side which are probably not in old domino code... Take a look at what is on IIS which should be at right level OK, thanks :) -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: Well so far I've taken the old Domino Connector source and built it against the latest version of ajp13/jk (as of last week). Should that not 'just work'? BTW, we're now focusing on jk2, so what about working on jk2 for domino. The requirement will be having apr for your box (windows ?) Yes, I noticed something to that effect. I'm going to spend another day trying to get this to work because we've got an urgent requirement for it. Once I get my immediate problem solved I'll have a look at jk2. I'm looking to support Windows, Linux and any other platform that supports Domino ultimately but I guess apr availability isn't going to be a problem on any Domino platform. -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: I'm trying to bring the Domino connector up to date with the latest codebase and I'm running into a problem with connection reuse. The characteristic is that with connection reuse working normally the first message from Domino to Tomcat round trips without any problem but a second message repeatedly re-sends without getting any response. If I modify ajp_send_request() not to recycle connections everything works fine. I'm going to investigate more but I'm wondering if those symptoms ring a bell with anyone... FWIW it's the same whether I configure Tomcat to listen with a Coyote connector or the old AJP 1.3 one. Thanks :) mod_jk 1.2.5 ? - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: mod_jk 1.2.5 ? 1.2.6 - the latest code from the CVS yesterday. -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: I'm trying to bring the Domino connector up to date with the latest codebase and I'm running into a problem with connection reuse. The characteristic is that with connection reuse working normally the first message from Domino to Tomcat round trips without any problem but a second message repeatedly re-sends without getting any response. If I modify ajp_send_request() not to recycle connections everything works fine. I'm going to investigate more but I'm wondering if those symptoms ring a bell with anyone... FWIW it's the same whether I configure Tomcat to listen with a Coyote connector or the old AJP 1.3 one. Could you get an ethereal dump ? - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: Could you get an ethereal dump ? That mail coincided exactly with my trying to remember whether I had a network tracing tool installed on this PC - and fortunately I do :) Ethereal dump coming up. -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Andy Armstrong wrote: Henri Gomez wrote: Could you get an ethereal dump ? That mail coincided exactly with my trying to remember whether I had a network tracing tool installed on this PC - and fortunately I do :) Ethereal dump coming up. In such case ethereal is allways your friend. I'll have to leave (Europe GMT end of day). - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: JK/ajp13 connection reuse problem
Henri Gomez wrote: I'll have to leave (Europe GMT end of day). UK time here. Still trying to get Ethereal to trace local (intra machine) traffic. I'm not around tomorrow so it may be Monday before there's anything to dissect. -- Andy Armstrong, Tagish - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]