=== menn0_ is now known as menn0-school-run === menn0-school-run is now known as menn0 [08:13] erik_lonroth_: any news? [08:47] kjackal: a chap called Brian Mullan said I should speak to you! ;) [09:09] hey magicaltrout, Brian is a wise man! [09:09] what about? [09:27] ha, he hooked up with me on linkedin kjackal and said he'd seen the questions about dcos and lxd [09:29] magicaltrout: wow, I must be more carefull on what write on the web.... [11:00] Hello there. I have a ha-juju env, and i lost 2 nodes one is still available, but that database is giving some locking issues atm. I Think i can fix that, but am i able to remove/repair the cluster or bring it back to just one node again? === BlackDex_ is now known as BlackDex [11:02] hi, I am not able to download spectrum scale manager charm [11:02] can anyone tell me what's the issue [11:03] here is the link : https://jujucharms.com/u/ibmcharmers/ibm-spectrum-scale-manager/13 [11:04] digvijay2016: what happens if you try to deploy it? [11:04] also what does `juju deploy --debug cs:~ibmcharmers/ibm-spectrum-scale-manager-13` tell you? [11:08] @BlackDex : here is the output http://paste.openstack.org/show/613264/ === bloodearnest_ is now known as bloodearnest [11:11] oke juju status should now show you that the app is there [11:11] only no unit/machine defined yet for it [11:13] but why I am not able to download the zip from UI? [11:15] i dont know [11:15] maybe you can download it using the charm-tools [11:17] digvijay2016: `charm-pull-source cs:~ibmcharmers/ibm-spectrum-scale-manager-13` did it for me :) [11:18] `sudo apt install charm-tools` to install those tools [11:18] let me try [11:22] working :) thanks. [11:22] yw :) [11:40] BlackDex: hmm, not really a "disable-ha" command. You can run with the warnings. If you can get it to a stable state perhaps you can migrate the models? (what version of juju is this?) [11:40] juju 1.25 [11:41] and juju dusable-ha is nice, but if there are no controllers to talk to ;) [11:43] BlackDex: heh, fair enough and yea 1.25 is going to be tough... [11:43] BlackDex: so if you can get the db back and have the one remaining node speak back to you you can reuse enable-ha I believe to bring up new controller nodes [11:44] i think i have mongo alive again [11:44] BlackDex: you might need to make sure to update the environments.yaml and in 1.25 there was some cache file for the controllers/models that had the list of IPs connected and the like [11:44] trying to check if i can remove the replset via mongo it self [11:44] BlackDex: so there should be some state in juju tracking the IPs of the nodes and such that might cause at least errors but hopefully can ignore for now [11:45] oke thanks for the tip [11:45] ill check if i can find that :) [11:45] First fix the db [11:46] kjackal: I'll have a look once I'm back home. I'm at work now discussing how we can ramp up our workplan towards juju, maas and and development of charms. We will need education here at some later stage specific to juju development. We are prepared to pay for good education. [11:46] BlackDex: yea, have to check the .jenv files to make sure the values in there are ok post-recovery [11:46] oke [11:47] do you maybe know where the user/pass is stored for the mongodb? [11:47] i need to auth but i can't find it [11:50] BlackDex: hmm, no. Sorry. It changed a bunch since 1.25 and I've no idea. [11:59] rick_h: found it: https://github.com/juju/juju/wiki/Login-into-MongoDB [11:59] well i can't connect it seems, but that is the right procedure [12:04] connected :). Now lets see if i can fix it all [12:15] erik_lonroth_: there are partnership programs that (useed to) include "charm schools" https://partners.ubuntu.com/programmes/charm [12:15] erik_lonroth_: I am not sure if charm schools are still included, I guess they are since its on the website [12:31] Thanx for pointing to that! [12:43] BlackDex: go go go! :) [12:44] msg lazyPower ping-a-do [12:44] bah [13:33] rick_h: pong === mup_ is now known as mup [16:33] reactive bash question: http://paste.ubuntu.com/24917986/ on initial install, *both* install_app() and config_foo() are executed, presumably because the config.changed.foo state is initially true. how can i alter config_foo() so it only gets executed when the user actually runs a "juju config app foo=bar"? [16:34] cory_fu: bash me brain smarts! ^ [16:40] kwmonroe: Does the foo config option have a default value that is not an empty string? [16:40] yes cory_fu, it's a true boolean by default [17:25] o/ juju world! [17:27] \o Budgie^Smore [17:32] The Juju Show #15 from the Canonical Offices in London in 30min! [17:32] kwmonroe: tvansteenburgh marcoceppi lazyPower bdx and others that might be interested in joining ^ [17:49] has anyone ran into the "Missing implementation for interface role" when building charms? - http://paste.ubuntu.com/24918410/ [17:57] Juju Show url to join the hangout: https://hangouts.google.com/hangouts/_/5w4l6x4l4jcgno5mke7ru252vye [17:57] Juju Show viewing url: https://www.youtube.com/watch?v=MRQYURC83zQ [18:15] juju docs recap: https://lists.ubuntu.com/archives/juju/2017-June/009132.html [18:37] hey .. sorry about the abrupt exit there .. looks as though I haven't been plugged in all morning :/ [18:39] bdx: lol all good. Your answer is in the video [18:40] excellent [18:41] * rick_h runs for food stuffs now [18:46] run? [18:47] hmmmmmmm k8s its been a while [18:47] * magicaltrout trys to figure out the DNS craziness [18:50] ^ ha [18:52] magicaltrout: slap deis atop that k8s cluster, point the fqdns at the loadbalancer and call it good :) [18:52] I feel like paas offerings are great for one off projects like client work [18:52] but when it comes to long running infrastructure, i prefer that tight control of a strictly declared manifest [18:53] I must say, letting deis handle the subdomain routing/proxying on a per application basis is quite nice [18:54] I'm sure it has its draw backs though [19:07] its no different than the ingress controller [19:07] only it assigns a random domain on a wildcard domain [19:17] * Budgie^Smore is catching up on The Juju Show [19:21] migrate is definitely nice! [19:22] yeahhhhh boiiii [19:26] hardly seems worth running k8s with only 2 machines though ;-) [19:32] Budgie^Smore: woot, it was for demo purposes :p [19:33] oh I see my confusion, I vaguely remember the k8s core bundle needing 4 machines [19:35] oh and you really should becareful when you stick those 2 fingers up with the back facing the person, especially in the UK ;-) [19:56] So I do have a question about charms since you have started working on CentOS ones, is there any thoughts on OS agnostic charms? [20:47] jupyter looks awesome [20:52] rick_h can I +1 for a Charm School Notebook? (or did I miss that being done already?) [21:11] Budgie^Smore: so, the general guidance there was to make good abstractions where it makes sense, but not to cram the kitchen sink into charms because it leads to messy solutions like poorly written cookbooks (as an example, i'm sorry chef fans not picking on ya) [21:11] where you wind up with abstracted abstractions that do 70% of what you want, but fall down because of vendor differences [21:12] LOL oh I get that [21:12] but there are many community members who are asking about this [21:12] so, i think there's a larger conversation that needs to happen between us, and the community, to make this streamlined. Like defining clear interfaces for these plumbing libs, and give users a consistent interface to bind their work to. [21:13] for example, i shouldn't care what series, i should be able to just say install_package('list, of, things, that, are, awesome') and get a consistent abstraction independent of implementor that does the right thing. installing my package, and thats all it does, anything else on top of that would be either a different abstraction, or manual python. [21:13] at least thats how i would expect it to work, maybe there are better patterns. [21:14] i'm not an expert in all things :) [21:16] can I discuss conjure-up in this channel? [21:17] Sure [21:17] axisys: ^ [21:17] conjure-up kubernetes fails with Unable to locate package kubernetes [21:17] stokachu: ^ [21:17] I am on ubuntu 16.0.2 LTS [21:18] axisys: what version of conjure-up [21:18] just ran sudo apt install conjure-up [21:18] conjure-up 0.1.2 [21:18] axisys: remove that package and follow the instructions from conjure-up.io [21:20] conjure-up 2.2.1 [21:20] Yea [21:21] stokachu: working .. sweet.. now I can follow the webinar.. thanks a lot! [21:22] axisys: np! [21:46] stokachu: we need to find and obliterate all sources pointing to the apt install method .... I'm sure you hear about it more then I do, but my users end up there too still somehow [21:47] bdx: yea i need to file a bug to have it removed in new releases and ive got a package that informs the user to use the snap [21:47] just need to go through the sru process [21:48] ahh, that would make sense [22:27] now with juju-2.2.0, all of the sudden I can't bootstrap aws [22:27] my vpc ids that worked before no longer work [22:27] I get these strange errors now ... writing up a bug in a few [22:28] gah [22:28] I'm getting successful bootstraps on other another vpc ... possibly just misconfiguration on my end [22:33] ahh .. did we lose the "--to 'subnet=subnet-3737373'" with 2.2 ? [22:34] for the bootstrap placement [22:34] ? [22:40] ahh I know whats wrong [22:40] there is no metadata in us-east-2! [22:40] how can I go about getting that metadata up there? [22:56] possibly I'm having an issue with dedicated vs default tenancy [23:19] Enter your Ubuntu SS credentials during conjure-up kubernets -> aws -> us-east-1 fails with ERROR cannot log into jimm.jujucharms.com .. but I can login to login.ubuntu.com with same credentials [23:19] what gives? [23:19] SSO*