On Wed, 1 Feb 2017 at 10:28 Andrew Lau <and...@andrewklau.com> wrote:
> If you use the cloud-provider nodes should be added and removed > automatically (at least this is the case we see with AWS). > > You would need to workout cert and config generation. What I would love to > see is the ability to trigger autoscaling by metrics from hawkular instead > of just raw usage in AWS. > > On Tue, 31 Jan 2017 at 03:20 David Conde <da...@donedeal.ie> wrote: > >> >> >> The only other approach I can think of is to have some sort of API to an >> openshift-ansible bastion host, that will run the "add node" playbook on >> the new node in response to a scale-up in the autoscaling group. >> That >> would be a nice simple approach, maybe something running on the bastion >> host that pulls from SQS and delegates to the ansible scripts. Is it >> currently easy and safe to remove a node using the ansible playbooks? >> >> >> On 30 January 2017 at 16:06, Seth Jennings <sjenn...@redhat.com> wrote: >> >> There is no on-list discussion of which I'm aware. >> >> Any solution would likely involve writing something like kubeadm for >> Kubernetes. That is, a mechanism for a node to provide a token/credentials >> to a cluster registration system and having the node certs/resources >> provisioned to join the cluster. >> >> The only other approach I can think of is to have some sort of API to an >> openshift-ansible bastion host, that will run the "add node" playbook on >> the new node in response to a scale-up in the autoscaling group. >> >> On Mon, Jan 30, 2017 at 9:38 AM, David Conde <da...@donedeal.ie> wrote: >> >> Hi Seth, >> >> Thanks for getting back to me, would the goal be to look at reusing what >> k8s has in beta at the moment? If you could point me at any discussions >> that have been had on it that would be great, maybe its something I could >> contribute back if I get it working. >> >> We are currently running on-prem but I'd love to have the ability to >> provision extra capacity on AWS when needed. >> >> Thanks, >> Dave >> >> >> On 30 January 2017 at 15:27, Seth Jennings <sjenn...@redhat.com> wrote: >> >> Upstream kube has this in beta. Origin doesn't support this right now. >> The real trick is joining nodes to the cluster. Currently, >> openshift-ansible has a playbook for joining nodes to an existing cluster >> but that pattern doesn't work really well for node autoscaling. >> >> We are looking at ways to do this, but there is no work done as of yet. >> >> On Mon, Jan 30, 2017 at 7:23 AM, David Conde <da...@donedeal.ie> wrote: >> >> Has any work been done on node autoscaling? >> >> I'd like to install an origin cluster on AWS with the ability to scale >> nodes up and down using something like an autoscaling group. >> >> >> Thanks, >> David Conde >> >> >> _______________________________________________ >> users mailing list >> users@lists.openshift.redhat.com >> http://lists.openshift.redhat.com/openshiftmm/listinfo/users >> >> >> >> >> >> _______________________________________________ >> users mailing list >> users@lists.openshift.redhat.com >> http://lists.openshift.redhat.com/openshiftmm/listinfo/users >> >
_______________________________________________ users mailing list users@lists.openshift.redhat.com http://lists.openshift.redhat.com/openshiftmm/listinfo/users
