[03:09] <kdavyd> Hi All. If this isn't the right channel, please point me the right way. Has anyone ever successfully managed to integrate MAAS node power controls and vSphere nodes via vCenter without going through a bunch of manual shenanigans and pinning VMs to individual vSphere hosts?
[03:12] <kdavyd> I see two ways - one is to hack on libvirt to allow it to talk to a vCenter via vpx:// without having to specify the ESX server name, just the cluster.
[03:13] <kdavyd> The other is to add vmware tools support into either the preseed or a custom commissioning script, and then get power controls to the VMs via wake-on-lan.
[03:14] <kdavyd> Neither option appears to be trivial, and google offers no help either.
[03:18] <kdavyd> What I'm trying to accomplish is a production-grade OpenStack HA deployment on top of an existing vSphere environment, with all OpenStack management nodes virtualized, and the only bare metal provisioned hosts being the actual compute and storage.
[04:27] <X-Rob> kdavyd: vmware have an openstack build that talks to vCenter
[04:27] <X-Rob> however, apparently it's poo.
[04:34] <kdavyd> X-Rob: I don't even want to down that path, partially for the reason you've described.
[04:34] <X-Rob> heh, yup.
[04:37] <kdavyd> (we've been through vCloud, and that didn't end up well)
[14:04] <dimitern> rvba, meeting?
[14:10] <rvba> dimitern: sorry I'm late…
[22:03] <jlondon> Howdy all. I'm trying Maas again after a couple years and am having some issues with the bootstrapping process. Some machines won't get past the initial pxe instantiation and just sit at the login prompt and the machines that do get their IPMI information added and then turned off... don't have ram/cpu info stored in maas. Any ideas? I've looked through the logs and didn't really see anything.
[22:22] <roadmr> jlondon: ram/cpu info is only collected when you commission them; the first contact with maas is called "enlistment" and it won't get that information
[22:22] <roadmr> jlondon: commission the nodes you want to use (there's a control for that), then reboot those nodes; they'll boot, then turn off again, but *then* you'll see their hardware information
[22:22] <roadmr> jlondon: once they're commissioned, you can start them or deploy workloads (e.g. with juju) on them
[22:23] <roadmr> jlondon: there's even a "commission and start node" button
[22:23] <jlondon> roadmr: Okay, understood. That part is working as expected then. Any ideas on troubleshooting the nodes which can't get past the 'maas-enlisting-node' login?
[22:23] <jlondon> I don't see anything in the logs about them and they never show up in the nodes list.
[22:24] <roadmr> jlondon: try switching VTs (alt-f1,2,3...) and see if another console shows what they're doing
[22:24] <roadmr> jlondon: if they got as far as showing the enlistment login, everything else should be ok
[22:25] <jlondon> All tty's show the enlisting login screen, no logs.
[22:26] <jlondon> Oh wait... tty7 has the dmesgs
[22:26] <jlondon> Last entry is 'Ign http://archive.ubuntu.com trusty-updates'
[22:28] <roadmr> jlondon: :D weird, is there some config difference between the failing nodes and the successful ones?
[22:28] <jlondon> roadmr: I seem to remember something like this happening the last time I tried Maas and it had something to do with that Maas sets up a proxy for the nodes or something like that?
[22:28] <roadmr> jlondon: network-wise, perhaps... a different switch?
[22:28] <roadmr> jlondon: yes, what's odd is, why do some nodes work while others don't
[22:28] <jlondon> roadmr: Nope. Same switches and I even just deleted one node which had been added okay and then tried adding it again and now it's hanging. Sooo... maybe a proxy issue?
[22:29] <jlondon> If so how do I disable the proxy?
[22:29] <roadmr> jlondon: That, I don't know :/ sorry.. since it's worked OK for me, I don't know how to disable it :(
[22:30] <roadmr> jlondon: you could post maas-related questions in askubuntu.com, maas people regularly check there for questions
[22:31] <jlondon> roadmr: Hrm, Okay thanks. I'll try to see if there's anything in the docs about it. I am now remembering that I had issues with the proxy the last time I tried this so.. Do you know how to login to the node before it has been found by maas? Maybe I can see the network info of the host from there.
[22:31] <roadmr> jlondon: hm no idea, particularly because I don't think the ephemeral image has a password :/
[22:32] <roadmr> jlondon: go to the console (alt-f7) right after the node boots, you'll be able to see cloud-init including network data there
[22:33] <jlondon> roadmr: Yeah saw the IP of one of the machines previously but couldn't get in with my ssh key or anything I tried. I think you might be right about no password though :(
[22:33] <jlondon> It's looking like it is a proxy issue now though, so I bet if I can disable it things will be okay.
[22:33] <roadmr> jlondon: right, they won't even be sshable until they're started I think :/ that's when cloud-init fetches the ssh keys from maas
[22:34] <roadmr> jlondon: have patience, someone here will know what to do. Just that I don't :) tomorrow morning you may have better luck, more people than now (almost end-of-day)
[22:35] <jlondon> :)
[22:35] <jlondon> Thanks for the help so far.
[22:35] <jlondon> 99% sure now that it's a proxy issue. The apt-get process or whatever is being run is going.. just very, very, very slowly.
[22:36] <roadmr> jlondon: np. Also sounds like wonky network card... hopefully you'll figure out the proxy and be able to rule this out.
[22:38] <jlondon> Yep. It's the proxy. Killed it and the node finished booting.
[22:38] <jlondon> Not certain why an issue I had two years ago is still an issue :|
[22:39] <roadmr> so weird...
[22:39] <roadmr> ok, need to log off. good night!
[22:39] <jlondon> Seeya~