Re: mod_jk - ajp hangs indefinitely on some requests
Roy McMorran wrote: Roy McMorran wrote: Despite my misgivings, 1.1.5 seems to have helped. It's run my test script for about 30 minutes now without problems (previously would have hung by minute 3). However! My dev system (which was already running 1.1.15) still exhibits the problem. It also has newer versions of Tomcat (6.0.18), Apache (2.2.10) and mod_jk (1.2.27). Correction. The combination of updating libtcnative to 1.1.15 _AND_ setting JkOptions +DisableReuse prevents both systems from hanging. I'm concerned about the performance implications of using this directive based on what I read at http://tomcat.apache.org/connectors-doc/reference/apache.html Don't be. With httpd and Tomcat on the same box (or even the same LAN) the overhead of establishing the TCP connection on every request is minimal. Whilst theoretically there would be a performance issue in practice you won't notice it and it is much more stable. I have heard some reports of people seeing better overall performance. The usual YMMV warning applies. The only way to know for sure is to test it. and I'm unclear as to why it's necessary. Can anyone educate me on this? Various causes, usually comes down to one end dropping the connection without the other one realising. HTTP keep-alive usually plays a part in this. Can also be related to connection time outs and when each end checks to see if the time out is expired. In short, disabling re-use removes all of these issues. Do I still have a problem with my configuration? You could probably tune the time out and keep alive settings to reduce the issue but just disabling reuse is usually the better option. Mark - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Mark Thomas schrieb: Roy McMorran wrote: Roy McMorran wrote: Despite my misgivings, 1.1.5 seems to have helped. It's run my test script for about 30 minutes now without problems (previously would have hung by minute 3). However! My dev system (which was already running 1.1.15) still exhibits the problem. It also has newer versions of Tomcat (6.0.18), Apache (2.2.10) and mod_jk (1.2.27). Correction. The combination of updating libtcnative to 1.1.15 _AND_ setting JkOptions +DisableReuse prevents both systems from hanging. I'm concerned about the performance implications of using this directive based on what I read at http://tomcat.apache.org/connectors-doc/reference/apache.html Don't be. With httpd and Tomcat on the same box (or even the same LAN) the overhead of establishing the TCP connection on every request is minimal. Whilst theoretically there would be a performance issue in practice you won't notice it and it is much more stable. I have heard some reports of people seeing better overall performance. The usual YMMV warning applies. The only way to know for sure is to test it. and I'm unclear as to why it's necessary. Can anyone educate me on this? Various causes, usually comes down to one end dropping the connection without the other one realising. HTTP keep-alive usually plays a part in this. Can also be related to connection time outs and when each end checks to see if the time out is expired. In short, disabling re-use removes all of these issues. Do I still have a problem with my configuration? You could probably tune the time out and keep alive settings to reduce the issue but just disabling reuse is usually the better option. Easier vs. better: If you setup sensible timeouts and use CPing/CPong as described on http://tomcat.apache.org/connectors-doc/generic_howto/timeouts.html There should be no such problems, even without DisableReuse. Regards, Rainer - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
mod_jk - ajp hangs indefinitely on some requests
Hello all, I'm seeing frequent cases where requests sent via mod_jk to Tomcat will hang indefinitely. The request will never return to Apache and the client will eventually time out. There are no error messages. In my test case I run wget against various URLs until the hang occurs (usually within 2-3 minutes). Here are the last few entries from my mod_jk.log file (JkLogLevel = trace): [Thu Dec 04 13:28:08.509 2008] [27224:1] [debug] ajp_connection_tcp_send_message::jk_ajp_common.c (911): 00c0FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 - [Thu Dec 04 13:28:08.510 2008] [27224:1] [trace] ajp_connection_tcp_send_message::jk_ajp_common.c (928): exit [Thu Dec 04 13:28:08.510 2008] [27224:1] [debug] ajp_send_request::jk_ajp_common.c (1325): (ajp13) request body to send 0 - request body to resend 0 [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_send_request::jk_ajp_common.c (1422): exit [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_get_reply::jk_ajp_common.c (1613): enter [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_connection_tcp_get_message::jk_ajp_common.c (953): enter Beyond this point there is no activity. There are never any error messages generated, not in mod_jk.log, catalina.out, nor in the Apache logs. There is a thread dump of Tomcat (after the hang begins) at: http://www.mdibl.org/~mcmorran/tomcat_thread_dump.txt This is a pretty simple setup, both Apache Tomcat are running on the same host, so there is no firewall issue. No load balancing, just one worker defined. workers.properties file: worker.list=ajp13 worker.ajp13.type=ajp13 worker.ajp13.host=localhost worker.ajp13.port=8009 worker.ajp13.socket_timeout=120 (have tried with and without the socket_timeout parameter - makes no difference) My server.xml file is at: http://www.mdibl.org/~mcmorran/server.xml.txt (I've tried various settings (or none) for connectionTimeout inside the AJP connector - makes no difference) mod_jk related parts of my httpd.conf file are at: http://www.mdibl.org/~mcmorran/httpd.conf.txt Version info: Server: Solaris 10 Apache: 2.2.6 mod_jk: 1.2.25 Tomcat: 6.0.14 libtcnative: 1.10 Java: 1.5.0_14 I'm running Tomcat under jsvc. The JVM gets 2G of memory (there are 8G total on the system) - I don't think there's any memory issue. I'm in test mode so I'm the only user; there's virtually no CPU load on the box. I've reached a dead end here -I'm hoping someone on the list may have some ideas. Thanks for any help you may be able to provide. Cheers, Roy -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Roy McMorran wrote: Hello all, I'm seeing frequent cases where requests sent via mod_jk to Tomcat will hang indefinitely. The request will never return to Apache and the client will eventually time out. There are no error messages. In my test case I run wget against various URLs until the hang occurs (usually within 2-3 minutes). Here are the last few entries from my mod_jk.log file (JkLogLevel = trace): [Thu Dec 04 13:28:08.509 2008] [27224:1] [debug] ajp_connection_tcp_send_message::jk_ajp_common.c (911): 00c0FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 - [Thu Dec 04 13:28:08.510 2008] [27224:1] [trace] ajp_connection_tcp_send_message::jk_ajp_common.c (928): exit [Thu Dec 04 13:28:08.510 2008] [27224:1] [debug] ajp_send_request::jk_ajp_common.c (1325): (ajp13) request body to send 0 - request body to resend 0 [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_send_request::jk_ajp_common.c (1422): exit [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_get_reply::jk_ajp_common.c (1613): enter [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_connection_tcp_get_message::jk_ajp_common.c (953): enter Beyond this point there is no activity. There are never any error messages generated, not in mod_jk.log, catalina.out, nor in the Apache logs. There is a thread dump of Tomcat (after the hang begins) at: http://www.mdibl.org/~mcmorran/tomcat_thread_dump.txt This is a pretty simple setup, both Apache Tomcat are running on the same host, so there is no firewall issue. No load balancing, just one worker defined. workers.properties file: worker.list=ajp13 worker.ajp13.type=ajp13 worker.ajp13.host=localhost worker.ajp13.port=8009 worker.ajp13.socket_timeout=120 (have tried with and without the socket_timeout parameter - makes no difference) My server.xml file is at: http://www.mdibl.org/~mcmorran/server.xml.txt (I've tried various settings (or none) for connectionTimeout inside the AJP connector - makes no difference) mod_jk related parts of my httpd.conf file are at: http://www.mdibl.org/~mcmorran/httpd.conf.txt Version info: Server: Solaris 10 Apache: 2.2.6 mod_jk: 1.2.25 Tomcat: 6.0.14 libtcnative: 1.10 Java: 1.5.0_14 I'm running Tomcat under jsvc. The JVM gets 2G of memory (there are 8G total on the system) - I don't think there's any memory issue. I'm in test mode so I'm the only user; there's virtually no CPU load on the box. I've reached a dead end here -I'm hoping someone on the list may have some ideas. Thanks for any help you may be able to provide. Have you tried: JkOptions +DisableReuse ? Mark - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Mark Thomas wrote: Roy McMorran wrote: Hello all, I'm seeing frequent cases where requests sent via mod_jk to Tomcat will hang indefinitely. The request will never return to Apache and the client will eventually time out. There are no error messages. In my test case I run wget against various URLs until the hang occurs (usually within 2-3 minutes). Here are the last few entries from my mod_jk.log file (JkLogLevel = trace): [Thu Dec 04 13:28:08.509 2008] [27224:1] [debug] ajp_connection_tcp_send_message::jk_ajp_common.c (911): 00c0FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 - [Thu Dec 04 13:28:08.510 2008] [27224:1] [trace] ajp_connection_tcp_send_message::jk_ajp_common.c (928): exit [Thu Dec 04 13:28:08.510 2008] [27224:1] [debug] ajp_send_request::jk_ajp_common.c (1325): (ajp13) request body to send 0 - request body to resend 0 [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_send_request::jk_ajp_common.c (1422): exit [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_get_reply::jk_ajp_common.c (1613): enter [Thu Dec 04 13:28:08.511 2008] [27224:1] [trace] ajp_connection_tcp_get_message::jk_ajp_common.c (953): enter Beyond this point there is no activity. There are never any error messages generated, not in mod_jk.log, catalina.out, nor in the Apache logs. There is a thread dump of Tomcat (after the hang begins) at: http://www.mdibl.org/~mcmorran/tomcat_thread_dump.txt This is a pretty simple setup, both Apache Tomcat are running on the same host, so there is no firewall issue. No load balancing, just one worker defined. workers.properties file: worker.list=ajp13 worker.ajp13.type=ajp13 worker.ajp13.host=localhost worker.ajp13.port=8009 worker.ajp13.socket_timeout=120 (have tried with and without the socket_timeout parameter - makes no difference) My server.xml file is at: http://www.mdibl.org/~mcmorran/server.xml.txt (I've tried various settings (or none) for connectionTimeout inside the AJP connector - makes no difference) mod_jk related parts of my httpd.conf file are at: http://www.mdibl.org/~mcmorran/httpd.conf.txt Version info: Server: Solaris 10 Apache: 2.2.6 mod_jk: 1.2.25 Tomcat: 6.0.14 libtcnative: 1.10 Java: 1.5.0_14 I'm running Tomcat under jsvc. The JVM gets 2G of memory (there are 8G total on the system) - I don't think there's any memory issue. I'm in test mode so I'm the only user; there's virtually no CPU load on the box. I've reached a dead end here -I'm hoping someone on the list may have some ideas. Thanks for any help you may be able to provide. Have you tried: JkOptions +DisableReuse ? Mark Hi Mark, thanks for the reply. I hadn't tried DisableReuse - the documentation at http://tomcat.apache.org/connectors-doc/reference/apache.html led me to believe that 1) it would likely be very detrimental to performance and 2) it probably doesn't apply my situation (no firewall, no network involved at all). Nevertheless, I gave it a go just now. Unfortunately it made no difference. Thanks, -r -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Roy McMorran wrote: libtcnative: 1.10 Sorry, should have seen this sooner. Try disabling it or using a newer version. Mark - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Mark Thomas wrote: Roy McMorran wrote: libtcnative: 1.10 Sorry, should have seen this sooner. Try disabling it or using a newer version. I'll give it a try but I'm not hopeful. You see I have two systems exhibiting this hang behavior. The other one (dev) has the latest libtcnative (1.1.15). And I have a third (the production box) also using 1.1.10 that is fine (it doesn't hang). I admit, it's strange. Thanks -r -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: mod_jk - ajp hangs indefinitely on some requests
How can i be removed from this awful blog? Date: Thu, 4 Dec 2008 15:50:30 -0500 From: [EMAIL PROTECTED] To: users@tomcat.apache.org Subject: Re: mod_jk - ajp hangs indefinitely on some requests Mark Thomas wrote: Roy McMorran wrote: libtcnative: 1.10 Sorry, should have seen this sooner. Try disabling it or using a newer version. I'll give it a try but I'm not hopeful. You see I have two systems exhibiting this hang behavior. The other one (dev) has the latest libtcnative (1.1.15). And I have a third (the production box) also using 1.1.10 that is fine (it doesn't hang). I admit, it's strange. Thanks -r -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] _ Send e-mail anywhere. No map, no compass. http://windowslive.com/Explore/hotmail?ocid=TXT_TAGLM_WL_hotmail_acq_anywhere_122008
Re: mod_jk - ajp hangs indefinitely on some requests
Doctor Khumalo wrote: How can i be removed from this awful blog? Just scroll down the page, Doctor, there's a link to do just that on every one of them. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Roy McMorran wrote: Mark Thomas wrote: Roy McMorran wrote: libtcnative: 1.10 Sorry, should have seen this sooner. Try disabling it or using a newer version. I'll give it a try but I'm not hopeful. You see I have two systems exhibiting this hang behavior. The other one (dev) has the latest libtcnative (1.1.15). And I have a third (the production box) also using 1.1.10 that is fine (it doesn't hang). I admit, it's strange. Despite my misgivings, 1.1.5 seems to have helped. It's run my test script for about 30 minutes now without problems (previously would have hung by minute 3). However! My dev system (which was already running 1.1.15) still exhibits the problem. It also has newer versions of Tomcat (6.0.18), Apache (2.2.10) and mod_jk (1.2.27). Looks like I'm not done yet. Thanks, -r -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: mod_jk - ajp hangs indefinitely on some requests
Roy McMorran wrote: Despite my misgivings, 1.1.5 seems to have helped. It's run my test script for about 30 minutes now without problems (previously would have hung by minute 3). However! My dev system (which was already running 1.1.15) still exhibits the problem. It also has newer versions of Tomcat (6.0.18), Apache (2.2.10) and mod_jk (1.2.27). Correction. The combination of updating libtcnative to 1.1.15 _AND_ setting JkOptions +DisableReuse prevents both systems from hanging. I'm concerned about the performance implications of using this directive based on what I read at http://tomcat.apache.org/connectors-doc/reference/apache.html and I'm unclear as to why it's necessary. Can anyone educate me on this? Do I still have a problem with my configuration? Thanks again, -r -- Roy McMorran Systems Administrator MDI Biological Laboratory [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
