mod_jk problem ...

[rafcio]

Hi,

we habe some problems with mod_jk. We have two JBoss instances (not in cluster). At the front we have Apache 1.3.29 with mod_jk 2.0.2 configured with sticky option. But sometimes we get errors:

[Wed Mar 17 10:17:20 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x935f288
[Wed Mar 17 10:17:20 2004] (error) [jk_worker_ajp13.c (389)]: ajp13.service() Error sending initial post -1 32 0
[Wed Mar 17 10:17:20 2004] ( info) [mod_jk2.c (465)]: mod_jk.translate(): uriMap /xml;jsessionid=5e2ies51i2o3i.omaasp1 lb:lb
[Wed Mar 17 10:17:20 2004] ( info) [mod_jk2.c (362)]: mod_jk.handler() per dir worker for 0x93436e8 0x935f288
[Wed Mar 17 10:17:20 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x935f288
[Wed Mar 17 10:17:31 2004] ( info) [mod_jk2.c (465)]: mod_jk.translate(): uriMap /plugin;jsessionid=5e2ies51i2o3i.omaasp1 lb:lb
[Wed Mar 17 10:17:31 2004] ( info) [mod_jk2.c (362)]: mod_jk.handler() per dir worker for 0x93436e8 0x9357ef8
[Wed Mar 17 10:17:31 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x9357ef8
[Wed Mar 17 10:17:31 2004] (error) [jk_channel_socket.c (557)]: channelSocket.receive(): Error receiving message body -1 0
[Wed Mar 17 10:17:31 2004] (error) [jk_workerEnv.c (482)]: workerEnv.processCallbacks() Error reading reply
[Wed Mar 17 10:17:31 2004] (error) [jk_worker_ajp13.c (416)]: ajp13.service() ajpGetReply recoverable error 21000
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (465)]: mod_jk.translate(): uriMap /xml;jsessionid=5e2ies51i2o3i.omaasp1 lb:lb
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (362)]: mod_jk.handler() per dir worker for 0x93436e8 0x935f288
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x935f288
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (465)]: mod_jk.translate(): uriMap /xml;jsessionid=5e2ies51i2o3i.omaasp1 lb:lb
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (362)]: mod_jk.handler() per dir worker for 0x93436e8 0x935f288
[Wed Mar 17 10:17:37 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x935f288
[Wed Mar 17 10:17:37 2004] (error) [jk_worker_ajp13.c (389)]: ajp13.service() Error sending initial post -1 32 0
[Wed Mar 17 10:17:38 2004] ( info) [mod_jk2.c (465)]: mod_jk.translate(): uriMap /xml;jsessionid=5e2ies51i2o3i.omaasp1 lb:lb
[Wed Mar 17 10:17:38 2004] ( info) [mod_jk2.c (362)]: mod_jk.handler() per dir worker for 0x93436e8 0x935f288
[Wed Mar 17 10:17:38 2004] ( info) [mod_jk2.c (405)]: modjk.handler() Calling lb:lb 0x935f288
[Wed Mar 17 10:17:38 2004] (error) [jk_worker_ajp13.c (389)]: ajp13.service() Error sending initial post -1 32 0

mod_jk is sending the requests to the wrong Jetty/JBoss instance. It#s look to be an mod_jk problem, but we didn't find any solution. If we configure mod_jk to send all requests to one instance, than it works fine.


Regards,
Rafal

[britt]

Hi, had the same problem

sometimes got the error
[jk_worker_ajp13.c (380)]: ajp13.service() Error sending initial post -1 0 0
the numbers at the end seemed to vary and also the message didn't occur all the time

I set the connectionTimeout attribute for the connectors in server.xml to
-1 and the errormessage is gone

setting this attribute to -1 also got rid of a message in stderr.log saying
Date org.apache.jk.common.ChannelSocket processConnection
INFO: connection timeout reached

hope this helps

regards
Britt