I maintain two WebObjects installation in China.
One of them is giving me problems - performance is slow. Our users
can't get connected, yet there are many sessions and much activity in
the logs.
I have 80 instances for 2 servers, they handle 10-30 sessions each.
Requests are completed
I would change from Load average to Round robin load balancing. It
seems to work better. Also, decrease number of instances and give each
instance more memory. Maybe 50 - 100 sessions per instance. Try to
configure 20 instances on one server, give each 200MB or ram.
Michael
Aspire Auctions
- Maximum Adaptor Threads is 512
Increasing this correlates to increasing incoming connections to
the HTTP server.
That seems waaay to high. If WO creates more than 16 or so worker
threads, it will have a hard time recovering without the users
getting timeout messages.
Max had some
On Mar 4, 2008, at 1:20 PM, Mike Schrag wrote:
- Maximum Adaptor Threads is 512
Increasing this correlates to increasing incoming connections to
the HTTP server.
That seems waaay to high. If WO creates more than 16 or so worker
threads, it will have a hard time recovering without the
I think it is because you are running headless but your app references
some client side Java stuff JavaEOApplication.framework?
On Mar 4, 2008, at 1:31 PM, Joshua Archer wrote:
I've run the instance at the command line with the following command:
On 4-Mar-08, at 4:56 PM, Art Isbell wrote:
On Mar 4, 2008, at 11:29 AM, Chuck Hill wrote:
On Mar 4, 2008, at 1:20 PM, Mike Schrag wrote:
Max had some really good info on all these values in the WO
scalability session from a few WWDC's ago ... He talked about how
the defaults are
On Mar 4, 2008, at 11:51 AM, Michael Kondratov wrote:
I would change from Load average to Round robin load balancing. It
seems to work better. Also, decrease number of instances and give
each instance more memory. Maybe 50 - 100 sessions per instance. Try
to configure 20 instances on one
On Mar 4, 2008, at 9:21 PM, Klaus Berkling wrote:
On Mar 4, 2008, at 11:51 AM, Michael Kondratov wrote:
I would change from Load average to Round robin load balancing. It
seems to work better. Also, decrease number of instances and give
each instance more memory. Maybe 50 - 100 sessions
On Mar 4, 2008, at 9:25 PM, Chuck Hill wrote:
Ok, changed the load balancing and also lowered the instances to
36, 20 on one 14 on the other server. One server has 4GB and the
other has 3GB. The ram was not set for the instances so I now set
it to 195M - the math worked out better this
On Mar 4, 2008, at 9:30 PM, Klaus Berkling wrote:
On Mar 4, 2008, at 9:25 PM, Chuck Hill wrote:
Ok, changed the load balancing and also lowered the instances to
36, 20 on one 14 on the other server. One server has 4GB and the
other has 3GB. The ram was not set for the instances so I
10 matches
Mail list logo
