I get rid of the balancer and set in apache configuration file:
JkMount /eltbinj/* ajp13w_01_01
With the 1.2.13 the result is obviously the same high speed (I unset the lang
for not italian (lucky) people):
...
13:13:00 (630.19 KB/s) - `/dev/null' saved [53048]
With the 1.2.26 the result is the same:
...
13:14:53 (7.35 KB/s) - `/dev/null' saved [53048]
So in conclusion, it is not a matter of load balancer.
Thanks for helping narrowing it down, also the information about it only
happening for dynamic content is interesting (maybe it has to do with
chunked encoding).
I could analyze the packet flow between Apache and Tomcat. If you set
JkLogLevel to trace (and if trace doesn't exist for 1.2.13 to debug
there), you get very long log output. If you do the wget test once with
each version and with the increased log level, I could have a look at
those log files. They will contain a full packet dump of the
communication and at least for 1.2.26 millisecond timestamps.
Nevertheless I'm still curious about your JK directives for Apache. I
guess JkMount is not everything ;)
Regards,
Rainer
---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]