tomcat-replication.jar with 4.1.18

2003-03-19 Thread Dennis Muhlestein
We deployed your latest version of tomcat-replication and are thus far
pleased with the results.  

Two minor issues:

1) Using the manager application to restart the context does not restart
the InMemorySessionManager all the way.  An exception is thrown stating
that the port is already taken. (using stop then start works as
expected).

2) Under load, when box B leaves the cluster, the boxes A and C
start to throw a whole bunch of ConnectionRefused Exceptions. 
Eventually, boxes A and C removed B from their list and the exceptions
stop.  I haven't been able to determine if the excess logging that
occured with the exceptions before the node was detected down negatively
affect the remaining boxes in the cluster or not.  This is with
debug=1 in the Manager config (server.xml).

Thanks for you work
-Dennis


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



RE: tomcat-replication.jar with 4.1.18

2003-03-19 Thread Filip Hanik
Hi Dennis,
thanks a bunch for the feedback, it helps a lot. I will clean out the error reporting,
so that you at most see only one message.

And I will look into what the Manager does when it does restart. One would think a 
restart is a stop and then a start, but maybe not :-)

Filip


 -Original Message-
 From: Dennis Muhlestein [mailto:  
 Sent: Wednesday, March 19, 2003 2:36 PM
 To: Filip Hanik
 Cc: [EMAIL PROTECTED]
 Subject: tomcat-replication.jar with 4.1.18
 
 
 We deployed your latest version of tomcat-replication and are thus far
 pleased with the results.  
 
 Two minor issues:
 
 1) Using the manager application to restart the context does 
 not restart
 the InMemorySessionManager all the way.  An exception is 
 thrown stating
 that the port is already taken. (using stop then start works as
 expected).
 
 2) Under load, when box B leaves the cluster, the boxes A and C
 start to throw a whole bunch of ConnectionRefused Exceptions. 
 Eventually, boxes A and C removed B from their list and the 
 exceptions
 stop.  I haven't been able to determine if the excess logging that
 occured with the exceptions before the node was detected down 
 negatively
 affect the remaining boxes in the cluster or not.  This is with
 debug=1 in the Manager config (server.xml).
 
 Thanks for you work
 -Dennis
 
 
 -
 To unsubscribe, e-mail: [EMAIL PROTECTED]
 For additional commands, e-mail: [EMAIL PROTECTED]
 
 

-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]