[14:28] <D4RKS1D3> hi i am trying to deploy masively machines in one cluster but is not working in paralell
[14:28] <D4RKS1D3> Someone the reason?
[14:28] <D4RKS1D3> It could be change in some place?
[16:37] <magicaltrout> hello folks if i want a purely manually provisioned setup
[16:37] <magicaltrout> what do i bootstrap?
[16:39] <rick_h> magicaltrout: https://jujucharms.com/docs/2.2/clouds-manual
[16:40] <magicaltrout> yeah just found it
[16:40] <magicaltrout> thanks
[16:41] <rick_h> magicaltrout: <3
[17:16] <magicaltrout> whats the deal with k8s and gpus?
[17:16] <magicaltrout> do i need to do anything special these days?
[17:20] <rick_h> magicaltrout: hmm, think there's a tutorial for that.
[17:20] <rick_h> magicaltrout: oh hmm, it's a link from the tutorial to sam's blog post https://medium.com/intuitionmachine/how-we-commoditized-gpus-for-kubernetes-7131f3e9231f
[17:21] <magicaltrout> yeah thats old school though
[17:21] <magicaltrout> i bumped in to chuck in IAD a few months ago who said run the edge snaps
[17:21] <magicaltrout> but i think they are mainline now
[17:21] <rick_h> magicaltrout: doh, maybe marcoceppi or tv<tab dammit not in the room> have more info
[17:22] <marcoceppi> magicaltrout: don't need to do much but just deploy
[17:22] <marcoceppi> 1.7.0 (and 1.6.0) will support 'em out of the box
[17:22] <magicaltrout> fair enough, i have a cluster coming up, i'll deploy a test container and see what happens thanks
[17:28] <magicaltrout> rick_h: did you get my response re thursdays vs tuesdays?
[17:29] <rick_h> magicaltrout: yes, checking with other folks that want to be in there about their schedules
[17:29] <magicaltrout> ah right
[17:29] <magicaltrout> no probs
[17:31] <magicaltrout> i got my openldap snap basics done today so i'm hoping to get that into a charm and connected to our ranger snap pretty sharpish to complete our base deployment
[17:31] <magicaltrout> I don't think we have too much more work get through, although juggling with about 1000 other things means it just ebbs and flows a bit
[19:43] <jhobbs> is there a way to use maas system_id instead of hostname as a placement directive with juju bootstrap? juju bootstrap maas2 --to <system_id> doesn't work as it expects a hostname
[19:44] <jhobbs> which is a bit awkward as maas doesn't use hostname's as primary keys for nodes, it uses system ids
[20:42] <magicaltrout> hey SaMnCo, good blog, the US Government is currently mining coins on your behalf! ;)
[20:48] <SaMnCo> magicaltrout: wait for the upcoming one then...
[20:48] <SaMnCo> Upcoming next week or so, I just need to rebuild my cluster post move
[20:49] <SaMnCo> Almost done :)
[20:49] <magicaltrout> nice
[20:50] <magicaltrout> I got asked to test GPU support via openstack on K8S, your blog was top of my google ;)
[20:55] <wpk> jhobbs: apparently not, file a bug please (maybe --to @instanceid)?
[20:55] <jhobbs> wpk thanks, bug filed https://bugs.launchpad.net/juju/+bug/1709995
[20:55] <mup> Bug #1709995: Can't use system id as a placement directive for MAAS <cdo-qa> <juju:New> <https://launchpad.net/bugs/1709995>
[21:18] <wpk> jhobbs: I'd do it but I'm on my vacations and I promised myself not to work ;)
[21:20] <jhobbs> wpk: aww get out! thanks for looking into it, though :)