=== tasdomas` is now known as tasdomas [04:07] Hi all, Is juju charms only for containers (means while deploying charm it is creating a container for that charm) [04:07] can we deploy our charm directly on a VM or Server? [04:12] ubabu: You can definitely deploy charms in VMs & bare metal servers [04:13] ubabu: (and containers) [04:15] blahdeblah: colud you please provide some refernce links about deploying charms on VMS & bare metal servers [04:16] ubabu: https://jujucharms.com/docs/stable/clouds-maas <-- bare metal [04:16] https://jujucharms.com/docs/stable/help-openstack or https://jujucharms.com/docs/stable/clouds-manual for VMs [04:17] ubabu: Also https://jujucharms.com/docs/stable/getting-started is pretty good [06:19] Hi all, how juju is getting OS images for containers? === mup_ is now known as mup [08:07] ubabu: That is taken care of by lxd/lxc :) === frankban|afk is now known as frankban === ahasenack is now known as Guest99073 === jamespag` is now known as jamespage === mup_ is now known as mup === Guest99073 is now known as ahasenack === ahasenack is now known as Guest76687 === perrito667 is now known as perrito666 === Guest76687 is now known as ahasenack === freyes__ is now known as freyes [15:25] mornin juju o/ [15:28] hm, how i can achieve that kind of scenario that only leader will fill the DB with init file? is there any decorator that I can use? [15:29] or any condition? [15:29] or @only_once will do the job? [15:31] Spaulding if you're using layer-leadership theres an @when('is-leader') state [15:32] mind you i have zero context to what you're attempting to do, i only see the last 3 messages [15:39] lazyPower: it's still a good tip! [15:40] i'll definitely check that [15:40] wow, it looks like what i needed! [15:41] thx lazyPower [15:41] np Spaulding happy to help [16:12] tvansteenburgh: Would you mind rebasing your local charm branch to get the get_cloud? I need that to test it in our workflow [16:25] cory_fu: yeah one sec [16:27] cory_fu: done [16:27] tvansteenburgh: Thanks! [16:29] cory_fu: i have one more small change to make and then i'll put up a PR for that branch [16:29] Awesome === bugg is now known as magicaltrout [16:38] lazyPower: I'm fixing the readme that bmullan pointed out on the list, you'll have a PR soon. [16:38] for the kube-core bundle [16:38] jcastro thanks! [16:38] jcastro there was a few i landed in the last day i put on cdk, that i managed to not put in kube-core [16:38] can i link you at those and collab push those in the same swoop? [16:39] sure [16:41] lazyPower: after we land in the upstream docs [16:41] I want to trim these pages down considerably [16:42] I don't want to have to update 3 READMEs [16:42] https://github.com/juju-solutions/bundle-kubernetes-core/pull/47 [16:43] Merged [16:43] jcastro https://github.com/juju-solutions/bundle-canonical-kubernetes/pull/170/files -- did you see this one sneak by on CDK? [16:44] that seems fine to me [16:50] jcastro, small comment on that PR [16:52] on it [16:53] https://github.com/juju-solutions/bundle-kubernetes-core/pull/48 [16:54] jcastro, \o/ [16:54] hmm, I wanted to tell him about updated CDK on the thread as well [16:56] stokachu: next week I'm going to do a new video for conjure-up/LXD [16:56] to replace the one we have now [16:56] jcastro, pimp [16:56] stokachu: think we can nail that ipv6/juju issue by then? [16:57] jcastro, yea so I updated conjure-up code already to add that Note in there about ipv6 [16:57] its already in the ppa [16:57] oh awesome [16:57] and i also handle if the user is not in the LXD group [16:58] i wanted to try and see about getting deis working with a conjure-up deployment, having some issues though [16:58] I have a plan for approaching deis [16:58] now that we spent a bunch of time with that deis guy yesterday at the sig ops meeting [16:59] nice, yea if you are able to manually get it working I can wrap that all up in the spell [16:59] I more want them to help for real, like a proper charm with integration [16:59] but I want to go with them with "this is a little hack we did, but we could be so awesome" [16:59] yea that would be nice [17:00] yep agreed [17:00] would be cool if on their website they just did 'conjure-up deis' [17:00] :D [17:01] too small [17:02] aww [17:02] "Using Canonical Kubernetes? You have it." [17:02] click here to pay us. [17:02] ah i see what you did there [17:04] thats slick [17:04] i like it :) [17:05] come on google, my stuff is passing, merge me [17:05] so _close_, then we won't have to have the docs spread out everywhere === holocron is now known as vmorris [18:47] cory_fu: whenever you get a chance: https://github.com/juju-solutions/matrix/pull/67 [18:49] petevg: Looks good at first glance, but should we move the full_stack test to follow that pattern? Or do we want to keep it where it is for code coverage? [18:50] cory_fu: leave it where it is for code coverage, I think. [18:50] Cool [18:54] petevg: In exchange: https://github.com/juju-solutions/matrix/pull/68 [18:57] cory_fu: looks good. merged. [18:59] petevg: Cool. The functional test stuff looks good, too. I'm running them now. About how long do they take? [19:00] cory_fu: they should take about as long as the full stack test -- the gating tests all skip actually deploying things. [19:04] cory_fu: it is spitting the matrix log to stdout, but I don't know how to make py.test/tox verbose enough to output it. You can tail /tmp//matrix.log if you get impatient :-) [19:15] petevg: It finished. Merged. Thanks [19:16] cory_fu: awesome. Thx :-) [20:38] cory_fu, i just noticed that nothing checks if cwr was successful -- it always pushes if push-to-channel was specified: https://github.com/juju-solutions/layer-cwr/blob/master/templates/BuildMyCharm/config.xml#L104 [20:39] shouldn't we only push on a successful cwr run? [20:39] kwmonroe: Correct [20:40] kwmonroe: Though, I think the job is set up to fail fast (set -e) so it shouldn't actually get to that part of the code if cwr sets its return value properly [20:44] cory_fu: it seems cwr doesn't think bundletester failures are fatal: (line 82): http://paste.ubuntu.com/23754237/ [20:44] kwmonroe: Yeah, that's why there's a high priority card on the board for that [20:46] sod matrix, don't test for failure, just assume it all works \o/ [20:46] :) [20:46] "With one simple change, we have made it to 100% passing tests" [20:46] woop [20:46] i assume you two are coming to gent? [20:49] bdx: you coming to gent this year to get off your face? [20:49] sorry i mean, talk about monitoring [20:49] magicaltrout: i think it's just petevg and kjackal from the big team going. it'll be too cold for cory_fu and myself. [20:49] loosers [20:49] fair enough [20:49] petevg's easier to wind up than cory_fu so thats acceptable [20:49] lol [20:50] fscking kjackal and he's endless talking about cars... might have to bring earplugs [20:51] well cory_fu you have no excuse to not do apachecon this year [20:51] well... unless mark doesn't want to pay your expenses which will be basically nothing [20:52] it's in miami in may, right magicaltrout? [20:52] correct [20:52] 18th i think [20:52] i'm there. just have the pesky task of getting a talk accepted. [20:53] aye, i need to submit some shit [20:53] ... i suppose i should check the cfp closing [20:53] see if they go for it [20:53] feb the something for the CFP [20:53] you've got a few weeks [20:53] phew - feb 11 deadline. [20:53] yup [20:53] Ugh. But Miami [20:53] will smith says it's a nice place [20:53] I lived there for a year, and that was one year too many [20:53] yeah never been. I believe most people in seville shrugged and sighed [20:54] It's probably great for visiting [20:54] well we only ever see a hotel and 1 block each side [20:54] As long as you want to spend a whole lot of money [20:54] i think i'm safe [20:55] i've been told JPL are planning a whole science track, dunno how likely that is to happen, but hopefully i'll get some apache - science - juju stuff accepted with this sparkler charm and some other bits [20:58] "Tracking down the crims on the dark web with Juju" [20:58] and get kwmonroe to say that in his british accent [20:58] it'll go down a treat [21:03] 2 right! === frankban is now known as frankban|afk [22:04] hello [22:04] if I have an issue with OpenStack deployed via juju and cloud-init/neurton, what team is my best bet to talk to? === spaok_ is now known as spaok [22:07] spaok: your best bet is probably to ask in #openstack-charms [22:08] thanks kwmonroe I'll try there [22:09] thats right kwmonroe you just pass the buck! [22:13] >< [22:14] heh, well kwmonroe is in that channel also :) [22:14] i wouldn't bother asking him... he'll tell you to use Hadoop [22:15] btw, anyone know what jujud's tipping point is for the number of nodes reporting in to it? [22:15] we seem to have killed ours [22:15] nope, but i believe you can change the default controller size if its really load that kills it [22:21] magicaltrout: we have an open issue with canonical about it, but was just curious [22:23] we've bumped the resources for the VM to like 64 cores and 128GB mem and juju can still take 10 min to respond to a status [22:23] magicaltrout: :) i clearly didn't /part #openstack-charms fast enough [22:23] spaok: what juju version? [22:23] and how long has the controller been up? [22:23] 2.0.0-xenial-amd64 [22:24] 13 days [22:24] we usually endup rebooting it, waiting 5min or so for it to settle and it responds for a bit [22:25] 51092 root 20 0 40.449g 2.695g 48732 S 252.5 2.1 37318:27 jujud [22:25] considering how many cores there are, 252% isn't too bad [22:25] but on reboots it will max all the cores [22:25] spaok: i had trouble with an azure controller that would become unusable after a couple weeks (https://bugs.launchpad.net/juju/+bug/1636634). i upgraded to 2.1-beta and it's been smooth sailing for the last 3 weeks (fingers still crossed). [22:25] Bug #1636634: azure controller becomes unusable after a few days [22:25] jujud is very core hungry [22:26] so 2.1-beta3 works for me, but i think 2.0.x had some mem improvements (according to comment 5 in the above bug). maybe worth an upgrade to 2.0.2 if you can. [22:27] ya, I have 2.0.2 in my other environment [22:27] the one with the neutron issues [22:27] does that one fair (fare?) any better? also, which word is correct in <-- that context? [22:27] well different scale [22:28] only has a 50 or so servers, the one with the issue recently had 500 servers added to two different models [22:29] kwmonroe: Yogi bear 'as many different meanings as an adjective, adverb, and a noan. it most commonly means just and unbiased, pleasin', crystal, and Billie Jean, or a public exhibition event. Grey Mare can be used verb and a noan. as a verb, it means ter Scapa Fla, get along, or succeed. [22:30] or [22:30] Fair has many different meanings as an adjective, adverb, and a noun. It most commonly means just and unbiased, pleasing, clear, and clean, or a public exhibition event. Fare can be used verb and a noun. As a verb, it means to go, get along, or succeed. if you don't understand [22:30] hah [22:31] where did the yogi bear come from? [22:31] Fair == Yogi bear [22:31] excellent. i fare well now. [22:31] in cockney [22:31] oh [22:31] hah [22:31] I'm like wtf [22:32] me too and i'm only 100 miles from london [22:32] but when i get bored I like to annoy kwmonroe [22:32] I'm in California, so yogi bear is a cartoon to me [22:32] you are bored a lot [22:32] http://www.whoohoo.co.uk/main.asp mostly by translating stuff on this website [22:32] kwmonroe: i work in IT... its acceptable [22:32] i think you mean exceptable [22:33] plus i work from like 9am to 12+ most days, so my down time is normally annoying you [22:33] or petevg if i'm feeling nice [22:33] I'm touched that you care. [22:33] lol [22:34] sooo petevg you're coming to gent? [22:34] ayup. Got the flight booked and everything. [22:34] blimey, ive not even looked at the train yet [22:35] * kwmonroe jots down 'blimey' for my accent antics [22:35] I'm bringing the family along, so I had to do more logistics. [22:35] jesus petevg what did i tell you about bringing family along to drinking events....? [22:35] sorry i mean work [22:35] Heh. [22:35] anyway, same applies [22:35] is it punishment for not buying the cigs in the duty free? [22:36] They know that I may be scarce for a couple of days. [22:36] Something like that. :-p [22:36] lol [22:36] or so the others can buy kinder egss.... legally! [22:36] \o/ [22:36] Exactly. :-) [22:37] make sure you know theh heimlich maneuver [22:37] I think that the kidlet will manage not to eat the toy ... [22:38] We'll brush up just in case. [22:38] *face palm* famous last words [22:38] :-p === scuttle` is now known as scuttle|afk [22:45] cory_fu: it just dawned on me that we're having bash shell out to cat a function that is then executed by python3... in the middle of a bash command hidden in a jenkins job xml. this is like, awesome. https://github.com/juju-solutions/layer-cwr/blob/master/templates/BuildMyCharm/config.xml#L96 [22:45] kwmonroe: Awesomely terrible, yeah. [22:46] whatever. the only good python3 is the python3 nested in bash. amirite magicaltrout?!? [22:46] kwmonroe: That would also be good to break out into a script. Those jobs really need to be broken up so that the XML just contains the info specific to the job and then calls out to the common code [22:46] that is the best python3 [22:46] simply the best..... better than all the rest.... better than.... [22:47] go [22:47] aye [22:47] rust [22:47] dart \o/ [22:47] everything should be written in dart [23:04] kwmonroe: Quick PR for you: https://github.com/juju-solutions/layer-cwr/pull/29 [23:05] gotta wait on travis [23:05] kwmonroe: Travis doesn't run on the charm [23:05] then we're gonna be waiting a while [23:05] lol [23:07] hey! that's nice. that would have caught a jclient.create_job that bit me earlier. merged