tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Betts, Chuck" <Chuck_Be...@intuit.com>
Subject Hardware Server failure with mod_jk
Date Wed, 09 Feb 2005 23:12:31 GMT
Hi,

 

We are doing research with Apache+mod_jk+Tomcat testing the failover
behavior when a hardware server failure occurs.  For example, the web server
loses power, a network cable becomes unplugged, etc.  We have Apache running
on a Web tier and Tomcat running on a separate App tier server.  If we
remove a network cable from the web to app tier, requests that mod_jk routes
to the failed app server just hang.  We tested the case where we simply kill
the Tomcat process, and in this case mod_jk properly senses this and
re-routes messages to other app servers in the cluster, so this problem
doesn't seem to occur for software failure.

 

I'm using the following versions of the software:

Apache 1.3.28

mod_jk 1.2.8

Tomcat 4.1.30

 

Here a (modified) snippet from our workers.properties file:

 

# worker to talk to tomcat on dmn1

worker.app_dmn1.type=ajp13

worker.app_dmn1.host=dmn1

worker.app_dmn1.port=4020

worker.app_dmn1.lbfactor=2.0

 

# worker to talk to tomcat on dmn2

worker.app_dmn2.type=ajp13

worker.app_dmn2.host=dmn2

worker.app_dmn2.port=4020

worker.app_dmn2.lbfactor=2.0

...

# worker to load balance between dmn1,dmn2

worker.app_lb.type=lb

worker.app_lb.balanced_workers=app_dmn1,app_dmn2 

 

We originally thought that adding the configuration setting
worker.app_dmnX.socket_timeout=10 would at least stop the requests from
hanging indefinitely, but that didn't seem to help.  Does anyone have
experience with this, and can recommend configuration settings to use to
handle this type of server failure?  Thanks.

 

-- Chuck

 


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message