On May 25, 2010, at 1:30 PM, Anjo Krank wrote:
~300 Transactions and 4-50 Sessions are, like, nothing if your
instance is up for more than 5 mins?
It was just a quick picture...
How could I have known :)
The distribution is the same after 100s of sessions and thousands of
transactions.
All it takes is one users looking at a lot of pages to skew these
figures.
That would skew the # of transactions, yes. But it should not skew
the creation of new sessions across instances. It is the session
allocation that is problematic, IMO.
Yeah, but from your figures, you might have simply had a few long
runners and a few quick exiters.
That does not explain the allocation of new sessions primarily to
instances 1, 3 and 5.
Can you try "ab"? This should show things with more accurate numbers.
ab -n 500 http://whereever
What would that show relevant to the woadaptor load balancer choosing
instances for requests without a woinst cookie or number in the URL?
I'm confused.
Apart from that, I dimly remember the host and port setting also
influenced the load distribution.
For Round Robin, it follows the instance order of configuration.
If you configure all the instances on Server A, then the ones on
Server B, then C, it will load up Server A first. You need to
interleave the configurations. Is that what you were thinking of?
Yeah right. One trick was to number the instances not <host><port>
but <port><host>.
Interesting idea. I thought it went by instance number.
Chuck
--
Chuck Hill Senior Consultant / VP Development
Practical WebObjects - for developers who want to increase their
overall knowledge of WebObjects or who are trying to solve specific
problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com
This email sent to [email protected]
