Re: loadbalance removes nodes one at a time

From: Rajat Banerjee <no email>
Date: Wed, 21 Oct 2015 18:28:03 -0400

Hi Ryan,
No real reason, except that Justin added remove_nodes more than a year
after I wrote the LB:

https://github.com/jtriley/StarCluster/blame/develop/starcluster/cluster.py#L200

So the loadbalancer uses remove_node() instead of remove_nodes() plural.
Me/You/Anyone else could modify and test the load balancer with the new
function, it should not be too difficult. I would be interested in adding
it but don't have a lot of spare time right now.

Best,
Raj

On Wed, Oct 21, 2015 at 3:11 PM, Ryan G <ngsbioinformatics_at_gmail.com> wrote:

> I'm using loadbalance to shrink my cluster after submitting a bunch of
> jobs. I noticed that StarClustser may identify a group of nodes that can
> be shutdown. It then shuts them down one by one by running each plugin for
> every node. Why not run the plugins once and remove the group of nodes at
> the same time? This seems a bit inefficient.
>
> _______________________________________________
> StarCluster mailing list
> StarCluster_at_mit.edu
> http://mailman.mit.edu/mailman/listinfo/starcluster
>
>
Received on Wed Oct 21 2015 - 18:28:24 EDT