On 08.06.2010 16:45, Mangold, Daniel wrote:
On 07.06.2010 20:22, Mangold, Daniel wrote:
First: sorry, it looks like at least half of my previous mail was truncated for 
whatever reason. The attachment did not go through as well. And it's true, the 
pasted log file above is not complete.
However, this is now my current configuration of workers.properties which seems 
to work.

worker.list=balancer,status

# DEFAULT CONFIG FOR WORKERS
worker.default.host=localhost
worker.default.type=ajp13
worker.default.socket_connect_timeout=5000
worker.default.socket_keepalive=true
worker.default.connection_pool_minsize=16
worker.default.connection_pool_size=1024
worker.default.connection_pool_timeout=3000
worker.default.reply_timeout=300000
# disable retries, whenever a part of the request was successfully send to the 
backend
worker.template.recovery_options=3

# Define Node1
worker.worker1.reference=worker.default
worker.worker1.port=8033

# Define Node2
worker.worker2.reference=worker.default
worker.worker2.port=8044

# Load balancing behaviour
worker.balancer.type=lb
worker.balancer.balance_workers=worker1,worker2

# Load balancing method can be [R]equest, [S]ession, [T]raffic, or [B]usyness
worker.balancer.method=S

worker.balancer.sticky_session=true
#worker.balancer.sticky_session_force=true
worker.balancer.max_reply_timeouts=10

# Status worker for managing load balancer
worker.status.type=status



Well...after trying different things, it seems that the problem was the force 
mode for sticky sessions. The Tomcat webapp requires sticky sessions for load 
balancing, otherwise it won't work. So this works fine now:
worker.balancer.sticky_session=true
#worker.balancer.sticky_session_force=true

When uncommenting the sticky_session_force, I always get the '503 service 
temporarily unavailable' message after the second click. If I read the log 
messages right, the reason I that mod_jk could not establish establish the 
connection to any of the Tomcat instances.

For a while I was desperate enough to try load balancing with 
isapi_redirect-1.2.30 on IIS instead of Apache web server. It behaves in the 
same way when I use the sticky_session_force property (service unavailable 
page). On the other hand, when commenting the sticky_session_force there, I had 
another problem. My guessing is that with IIS and isapi-redirect, the 
sticky_session property did not work at all. But maybe I misconfigured 
IIS...I'm not really familiar with it.

Are there any known issues with sticky_session on Apache Webserver or IIS?

Most of the code is the same for IIS and Apache, especially all the load balancing stuff. So no difference to expect.

There are no known issues around session stickyness. From what I remember in your incomplete log snippet, your JSESSIONID did not contain a worker route. In order to make session stickyness work, you need to set jvmRoute in server.xml of yur tomcat to the same value as the name of the worker pointing to that Tomcat ("worker1" resp. "worker2"). Tomcat will append the value of the jvmRoute at the end of each session id, separated with a dot. mod_jk will find the route there and then look up the correct worker by name.

Look at

http://tomcat.apache.org/connectors-doc/generic_howto/loadbalancers.html

for instance the second block in red.

Regards,

Rainer

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org

Reply via email to