Hi Val,

"Prematurely" in this case is a Node not fully initialized.

In a nutshell, Ignite is started, enters the "discovery set", but further
things must happen for it to be ready to do computation (take real load).
When the Grid is initially started, this is pretty easy to control -- we can
gate that at the load-balancer -- only allowing it to receive load when the
Grid is "complete", and the caches are primed. By allowing the health-check
to only return a 200 when the Grid can truly receive load.

But later, when the Grid is complete, a Node entering the Grid must still be
fully initialized before it can take load. But there is no easy way to say;
"I am ready for load" for the ComputeTask. It seems there really needs to
be. 

That said, I do not think the AdaptiveLoadBalancingSpi works in my
scenario??

Because we are using Affinity, and the docs seem to indicate that load
balancing and Affinity are mutually exclusive.


> Data Affinity
> Note that load balancing is triggered whenever your jobs are not
> collocated with data or have no real preference on which node to execute.
> If Collocation Of Compute and Data is used, then data affinity takes
> priority over load balancing.

A further question:

Could you please confirm that:  Affinity::mapKeysToNodes() returns ONLY
PRIMARY Nodes ??
The javadoc is unclear there.

And that I would need to use Affinity::mapKeyToPrimaryAndBackups() if I want
to provide alternate Nodes when I want to provide alternate Nodes.

Thanks,
-- Chris 



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Reply via email to