[03:45] <xavpaice> hey there, is there a way to get logs for the openstack API calls the controller is making during `juju add-machine --constraints "instance-type=kubernetes-xxlarge zones=k8s-rack3"` ?
[03:45] <xavpaice> I need to know what it's requesting, and which openstack user it's using to confirm we have the quotas right
[05:01] <xavpaice> I tried increasing the logging to DEBUG, but so far it just tells me that "ERROR juju.worker.provisioner provisioner_task.go:985 cannot start instance for machine "41": suitable availability zone for machine 41 not found".  There's two machines each for k8s-rack1 and k8s-rack2, but only one for k8s-rack3, and we already have machines on that AZ.  Is that the issue?  Capacity is fine, but maybe some anti-affinity thing going on?
[08:14] <Hybrid512> o/
[10:39] <gnuoy> stickupkid, thanks for you help with getting the spaces patches landed.
[10:40] <stickupkid> gnuoy, more than happy to, anytime!
=== dannf is now known as dannf`
[16:49] <teldredge> Does Juju select a MaaS machine entirely at random or is there a way to have it choose a machine closests to its CPU\RAM\HDD constraints?
[16:50] <stickupkid> teldredge, https://juju.is/docs/constraints
[16:53] <teldredge> I understand it must choose a machine that satisfies the minimum constraints. I'm wondering if it's random from there. If the constraint is 4GB and it has an 8 and a 16 available- does it choose at random or select the 8 for being closer?
[17:28] <stickupkid> teldredge, It's what ever maas gives us back and we use that order, so I would say first one that meets the constraints
[17:28] <stickupkid> teldredge, if you want the 16GB one, then selecting anything higher than 8 will do the job
[17:32] <teldredge> Gotcha, I was hoping it would choose the closest match. I'm using Proxmox so I have a bunch of VMs MaaS can grab when it needs them so I was hoping to make it use the least resources needed, but I probably just need to use tags.
[18:30] <teldredge> Curious if anyone has seen any projects deploying kubevirt with juju
=== gurmble is now known as grumble
[20:45] <teldredge> If I bootstrap a juju-controller and use it to deploy a kubernetes cluster, do I have to deploy another juju-controller running in the cluster?
[20:45] <teldredge> Or can I use the original juju-controller to control the Maas 'cloud' and the k8s cluster?
[21:09] <thumper> you should be able to do the later
[21:09] <thumper> adding the k8s cluster to the controller