Hi,

>There does not seem to be any documentation for Tomcat in a
multihosting
>environment.  The main things which are lacking are the results of
setting
>certain parameters too high or too low.

I've also mentioned MANY MANY MANY times why such documentation can be
worse than useless, and actually misleading: it's specific to the
application and the environment being documented.  These characteristics
greatly impact performance and the specific setup is just that:
specific.  Set it up and test/tune repeatedly until you're happy.  Then
if you feel like it post your findings so that other people who think
they're a silver bullet might "benefit."

>Ex. MaxProcessors (Tomcat 4.1.27)  - we have an issue where one site
will
>creep up to 107 thread and be so slow that we have to restart the
service
>(JVM Instance.)  Default value is 75 - how does it even get up to 107?
We

75 is the number of processor threads.  There are many others in the
JVM, some related to tomcat and some that are not.  You can use a
profiler, a simple little utility (like the SystemThreadList.java class
I've posted to the list numerous times), or a ctrl-break thread dump to
see what all the threads are.

>have been constantly modifiying this along with accept count and
monitoring
>trying to figure the best fit.  However, I don't know if this is a good
>value to be modifying when you have 20+ sites.

You're going through the right process, assuming you're doing it in a
test environment.

Also please be careful in your usage of "site," "instance," "web
application," "context," and related terms.

>1) Is Tomcat Standalone (4.1.27 or higher) still a viable
implementation in
>a multi-instance scenario?

Yes, we've had such setups in production for years now.

>2) What are the ramifications of modifing these parameters in a
>multi-instance scenario?

They impact performance directly.  You have to test modifications out in
a test environment before employing them in production.

>3) Is anyone willing to join a create a discussion group on this topic
or
>should we keep posting our thoughts/concerns here?

Stay here: this is a central source for information about tomcat, it's
widely archived and referenced.

>So far the testing of standalone tomcat proves to be ok, however, I am
>weary
>of changing production because test environment and load tests almost
never
>provide the same results as real life :)

If you have little faith in your tests, that's unfortunate.  As long as
they simulate why you think is a realistic load, you'll not only be
fine, but you'll be covered (to higher ups) to a good extent if things
go wrong.

Yoav



This e-mail, including any attachments, is a confidential business communication, and 
may contain information that is confidential, proprietary and/or privileged.  This 
e-mail is intended only for the individual(s) to whom it is addressed, and may not be 
saved, copied, printed, disclosed or used by anyone else.  If you are not the(an) 
intended recipient, please immediately delete this e-mail from your computer system 
and notify the sender.  Thank you.


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

Reply via email to