Stephanie Wullbieter schrieb:
Please have a look at this reference worker timeouts:
worker.vorlagedummy.port=8009
worker.vorlagedummy.type=ajp13
worker.vorlagedummy.sticky_session=True
worker.vorlagedummy.socket_keepalive=True
worker.vorlagedummy.connect_timeout=50000
worker.vorlagedummy.prepost_timeout=50000
50 seconds is pretty long for a connect and a prepost timeout.
As you say, your network is unstable. If you want mod_jk to wait for
such a long time to finally succeed, 50 seconds will be OK, but this
will mean, that under high load a lot of requests might wait and block
httpd ressources. If you prefer mod_jk to throw and error and clear the
system whenever a long network latency shows up, reduce those two
timeouts to something like 10 seconds.
worker.vorlagedummy.connection_pool_timeout=30
This is pretty short. The timeout determines, how long we keep an idle
backend connection open for reuse. Usually it is more in the range of
300 seconds. You should also adjust connectionTimeout on the Tomcat AJP
connector to the same value multiplied by 1000 (it is in milliseconds on
the Tomcat side).
Some more info on the timepouts is contained in
http://tomcat.apache.org/connectors-doc/generic_howto/timeouts.html
We switch from mod_jk 1.2.5 to 1.2.23
If you are doing the upgrade now (which is good), I would suggest to go
directly to 1.2.26.
And with 1.2.5 we had a socket_timeout (now named connection_pool_timeout) of
30s and the other timouts are new. So first I want to set high timeouts because
our network is unstable with sometimes big latency (e.g. 20s) because the
firewall is sometimes overloaded.
No, there still is a socket_timeout. It is *not* the same as the
connection_pool_timeout.
Is this ok ?
See above :)
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]
