[00:01] k done : http://paste.ubuntu.com/23631252/ [00:02] teranet: what does MAAS say now? [00:02] 3 Ready 3 deployed [00:02] teranet: okay, so you've got 3 nodes doing something else? [00:03] nope [00:03] should I release those too ? [00:03] teranet: release those manually then [00:03] 1 I know had a hardware issue which I fixed [00:03] ok will do [00:04] ok now they are all ready [00:05] so now I should bootstap again ? [00:07] correct ? [00:12] ok bootstapping a controller again [00:12] teranet: yes, bootstrap and deploy again [00:13] for juju-gui can I only deploy the gui to make it work or does the rabbitmq server has to be deployed too? [00:13] and mysql sorry forgot that one LOL [00:14] teranet: what I would do, when all your maas machines are in a 'ready' state, is `juju add-machine -n 6` [00:14] teranet: and just make sure they all deploy successfully without any charms or bundles [00:14] hmm ok will do that [00:15] then once you verify that, the rest should fall into place [00:15] how do I add later more to it ?? does juju does it automaticly once maas say''s they are ready ? [00:15] teranet: yea, your `juju status` will show them as 'started' [00:16] if you succeed in ^, the rest will work given you have your maas<->juju<->openstack networking aligned [00:16] ok will check it [00:21] teranet: juju gui doesn't require anything [00:21] teranet: it's already deployed by default, just run `juju gui` command [00:24] ah ok cool [04:30] Hi everybody, I'm trying to use this bundle https://jujucharms.com/kubernetes-core/ but it's installing the latest version of kubernetes. How can I force it to use an older version? Thanks a lot! [04:48] skay: ta. I'll test and merge that and add some docs on the charm store publication workaround. [06:03] skay: released [06:28] kwmonroe: My issue on that is https://github.com/juju/charmstore-client/issues/103 === frankban|afk is now known as frankban === jamespag` is now known as jamespage [08:46] Good morning Juju world! === arosales_ is now known as arosales === mup_ is now known as mup [13:30] Hello, is it possible to auto-scale (regardless the infra) the worker node of a juju kubernetes deployment? [13:30] or do you have to create a custom solution that monitors kube workers load and decides whether to request for extra resources (lxd, VMs) using the API of the underlying infra? [13:33] frankban: ping [13:33] voidspace: hey [13:33] frankban: hey, hi [13:34] frankban: I have made a change to bundlechanges that I would appreciate you looking at [13:34] voidspace: sure [13:34] frankban: https://github.com/juju/bundlechanges/pull/29 [13:34] frankban: we need container placement to honour application constraints [13:34] frankban: this change implements that [13:34] frankban: but the bundlechanges package is new to me :-) [13:35] frankban: when you get a chance anyway [13:36] voidspace: looking [13:37] frankban: for context, this is the juju bug https://bugs.launchpad.net/juju/+bug/1626597 [13:37] Bug #1626597: Juju ignores constraints set in the bundle and deploys KVMs with default values <4010> [13:53] voidspace: reviewed [13:54] frankban: your analysis of the bug seems correct and passing constraint only rather than application is reasonable, however also [13:54] frankban: :-) [13:55] frankban: the new test, when the unit is located to "new", that can't be a container can it? [13:55] voidspace: yeah that's the kind of "take it or leave it" suggestion, the branch is good, happy to see that bug discovered [13:55] frankban: a unit can only be located on a container if placement is specified [13:55] frankban: i.e. I'm not entirely sure what you mean by "new" :-) [13:55] voidspace: it cannot, but since your change touches that it would be nice if that's exercised by a test anyway [13:56] voidspace: "new" is a special placement meaning new top level machine [13:56] frankban: right [13:57] frankban: is that an explicit placement? [13:57] frankban: ah, I see it in the code - I will try and work it out [13:57] voidspace: which is the default if no placement is specified, but for instance, IIRC, can be used in a multiple placement definition, like to: ["1", "new", "lxd:2"] or similar [13:57] frankban: ah, I see [13:57] frankban: understood [13:57] cool [13:58] frankban: I may be able to remove that change - let me check [13:58] voidspace: for instance https://github.com/voidspace/bundlechanges/blob/37e0752c3c530d1af168b3a2f90592dc9ce85549/changes_test.go#L286 [13:58] frankban: I added the change there too because I saw a ContainerType [13:59] frankban: yep [13:59] voidspace: I don't think that's a bad change, maybe that's required as well [13:59] frankban: right, it might actually be a different bug... [13:59] frankban: ok, I'll add a new test :-) [13:59] voidspace: ty [13:59] the new machine should honour application constraints as well [14:00] indeed === frankban is now known as frankban|afk === frankban|afk is now known as frankban [15:53] juju deploy /home/juser/temp/charm-ubuntu/builds/ubuntu monitor - 27 of those, 25 started, 2 are State:pending [15:54] it has been like that at least 5 hours. [16:00] frankban: I've added a new test for the case you suggested and made the change to explicitly pass constraints rather than the whole application spec [16:00] frankban: and I'm going to land the change [16:00] voidspace: cool thanks [16:00] frankban: thanks for your help [16:01] frankban: is it the usual $$magic$$ to trigger the landing bot on that branch as far as you know? [16:03] frankban: if there's no landing bot I might will have to ask you to land it, as I don't have write access to that repo [16:06] frankban: it's alright, I found the magic... [16:07] voidspace: sorry, on call, it's :shipit: probably [16:07] frankban: it is, and it's done - sorry for the noise [16:07] np [16:31] tvansteenburgh: I'm trying to see the delta between the latest and next-most-recent review rev of https://review.jujucharms.com/reviews/24 and it seems to include a bunch of stuff that was not specifically changed in the latest rev. Am I missing something? [16:33] * tvansteenburgh looks [16:36] tvansteenburgh: nm, I was reading the diff wrong. It's actually showing the right stuff [16:36] cory_fu: can you give an example of a something that's displayed but shouldn't be? like, name a file [16:36] oh okay [16:47] ok quick question on juju network ranges [16:48] I have my MAAS to use 10.5.x.x/24 but somehow now all of the sudden when I deployed juju charms those charms took 10.0.0.x IP's where do I can see and change that ? [17:22] teranet: that's odd, do you have multiple spaces configured? [17:24] no not yet [17:25] where could I see what spaces / ranges juju uses ? [17:25] teranet: it reads them from maas. You can use list-spaces to see what spaces it sees [17:25] teranet: and show-machine 0 to see details about the machine and what networks it's on [17:27] ok that only shows what I configured : http://paste.ubuntu.com/23634617/ [17:28] but below you can see my charm's somehow have also 10.0.0 in use ..... [17:28] teranet: adjust the number of the machine to the one that your charm is deployed on [17:28] can that be something within the charms ? even when I deployed the yaml none had any network setups in it [17:29] teranet: not really, the charms are handed a machine to run on so it's what maas has for network config and what you've done in Juju to specify deployment constraints [17:29] ok updated [17:30] teranet: when you update the pastebin the link url changes [17:30] teranet: so can't see any changes with the other url [17:30] http://paste.ubuntu.com/23634623/ [17:30] ups :-) there we go [17:33] teranet: does MAAS provide dhcp and is it providing dhcp on both subnets? [17:33] teranet: in Juju the containers should be getting their IP addresses from the MAAS dhcp server [17:35] MAAs provides DHCP on the default which is 10.5.100... [17:35] but only on eth0 [17:35] eth1 is reserved for public only === frankban is now known as frankban|afk === frankban|afk is now known as frankban === petevg_afk is now known as petevg === frankban is now known as frankban|afk [20:39] any idea ? === frankban|afk is now known as frankban === frankban is now known as frankban|afk [23:06] how do I tell juju2.0 to use a specific LXD image? [23:06] that I installed into LXD manually [23:26] or do I have no choice and I have to use simplestreams? (the environment this is going in is completely offline, with manually syncing of everythng)