[07:40] <stub> cory_fu: whoops, sorry.
[07:41] <stub> uploaded
[13:30] <elmaciej> Hello! Does anyone have expierience with installing mapr using juju
[13:50] <rick_h> Sorry elmaciej  not done it here. What charm are you using and what are you hitting?
[14:37] <SuneK> Hi, I'm trying to deploy the canonical kubernetes cluster through juju gui. I exported and modified the yaml file to add constraints for the workers, however these constraints were not met by the provisioned machines.
[14:38] <SuneK> (Of course I also reimported the updated yaml file :-)
[14:39] <SuneK> I know I could just add constraints to the machines, but as far as I understood adding the constraints to the applications, should ensure, that the application was deployed on machines with sufficient resources
[14:49] <SuneK> msg NickServ register 0G2YOU8ipDoh sune.kjaergaard@gmail.com
[14:49] <SuneK> lol
[14:53] <cory_fu> stub: Thanks.  Any chance we could automate the release process somehow so that they stay in sync?  Or perhaps give me and / or tvansteenburgh access on pypi to update it if it's missed?  We depend on the pypi version in conjure-up (we use pypi to bundle it into the snap)
[14:55] <stub> PPA gets built automatically, snap published to edge automatically but needs manual promotion to stable (by me, since snapstore doesn't have teams), pypi isn't automatic
[14:55] <stub> I can give you and Tim access to Pypi, sure
[14:56] <cory_fu> Thanks.  That'll at least take care of the case where one or two of us is unavailable
[14:57] <stub> It will lose gpg signing by my key, but I don't think anyone in the real world cares about that.
[14:58] <akshay__> Hi All, on my OCATA setup (openstack+my application) when I revert the snapshot to openstack only state and then try to re-deploy my application it is not allowing me to add relation and errors out as the relation being added already exists. Can someone please help me out.
[14:58] <stub> cory_fu: Is done
[15:00] <cory_fu> stub: Thanks!
[15:46] <kwmonroe> rats, i missed SuneK.  if that person comes back with the application vs machine contraint issue, it's bug 1676986.  workaround is that you need machine constraints to affect initial bundle deployment; app constraints alone won't do it.
[15:46] <mup> Bug #1676986: juju doesn't honor bundle application constraints (2.1.2) <constraints> <juju:Invalid> <https://launchpad.net/bugs/1676986>
[16:02] <magicaltrout> is there an answer you don't know kwmonroe .......
[16:09] <kwmonroe> magicaltrout: i don't know why you keep messing with my backscroll!
[17:16] <rick_h> kwmonroe: magicaltrout bdx hml externalreality_ and folks a heads up on juju show in 105min
[17:16] <hml> rick_h: :-) - what’s today’s topic?
[17:16] <hml> agprado: ^^
[17:16] <rick_h> hml: upgrading juju
[17:17] <rick_h> agprado: that's what I was looking for. I was looking in the G's trying to find his nick
[17:17] <rick_h> hml: IS has a cool tool for upgrading juju controllers and models I want to show off and get folks involved with
[18:20] <yosefrow_> @kwmonroe, is it possible yet to deploy a bundle with a manual cloud. iirc last time I tried that, it failed because juju could not use the manually added machines to deploy for some reason.
[18:22] <rick_h> yosefrow_: so the newer juju supports a flag on bundle deploy --use-existing that should work out for you
[18:23] <rick_h> oh damn too slow and wrong command https://jujucharms.com/docs/2.3/charms-bundles#recycling-machines
[18:24] <kwmonroe> heh
[18:24] <kwmonroe> he'll be back.  they always come back.
[18:26]  * rick_h adds that as a note to the juju show
[18:26] <rick_h> 30min! and then party time!
[18:34] <yosefrow> rick_h, wheres the party?
[18:35] <rick_h> yosefrow: juju show in 26min!
[18:35] <yosefrow> :o
[18:35] <yosefrow> where?
[18:35] <rick_h> yosefrow: and https://jujucharms.com/docs/2.3/charms-bundles#recycling-machines for ya
[18:35] <rick_h> yosefrow: https://www.youtube.com/watch?v=Pkbp4VK8-vo
[18:35] <rick_h> at the top of the hour
[18:35] <yosefrow> @rick_h, this was in answer to my question?
[18:36] <yosefrow> recycling
[18:36] <rick_h> yosefrow: yep, if I understand your question correctly
[18:36] <yosefrow> this looks amazing
[18:36] <yosefrow> its probably exactly what i need
[18:36] <yosefrow> wish i knew about that months ago
[18:37] <rick_h> yosefrow: well it's a month old I think in the last release or two
[18:37] <yosefrow> ah i heard about it then
[18:37] <rick_h> so yea, added a note to mention it on the juju show today in case it slipped by others
[18:37] <yosefrow> when i originally had the problem i complained to canonical and they said is a feature coming to the next release
[18:38] <yosefrow> @rick_h, specifically useful when deploying manual cloud
[18:38] <rick_h> yosefrow: definitely
[18:38] <rick_h> yosefrow: what machines are you running on then?
[18:38] <rick_h> yosefrow: is it something we don't support as a cloud or just machines lying around or ?
[18:38] <yosefrow> my use case is 3 nodes manually provided by a client who doesnt want me to access their hypervisor. Then bootstrapping to kvm node on management node and deploying k8s to the 3 manual nodes with a bundle
[18:39] <rick_h> yosefrow: oic
[18:39] <yosefrow> i fell short of deployment and was forced to use rancher
[18:39] <yosefrow> because juju insisted on ignoring my manually added machines
[18:39] <yosefrow> glad to see this fixed
[18:40] <rick_h> yosefrow: definitely
[18:40] <yosefrow> ill include it in my blog post about manual deployment with juju when i get around to it
[18:41] <rick_h> yosefrow: I'd love to see it when you get around to it
[18:41] <yosefrow> sure
[18:41] <yosefrow> you can see my previous posts and correct any glaring issues if you want
[18:42] <yosefrow> havent had many ppl review it and im wondering if my posts are clear enough
[18:42] <yosefrow> rick_h, do you work with maas much?
[18:42] <rick_h> yosefrow: I use it, I've got my own maas cluster at the house I use for testing and such but don't work on it
[18:42] <rick_h> yosefrow: but I'm a fan :)
[18:43] <yosefrow> cool
[18:43] <yosefrow> how deeply are you involved with juju?
[18:43] <rick_h> I think it really solves a cool problem that more folks have than realize (maas that is)
[18:43] <rick_h> I've worked all around it for a while. I'm pretty involved.
[18:43] <rick_h> (juju that is)
[18:45] <yosefrow> nice
[18:45] <yosefrow> you write code for juju?
[18:46] <rick_h> I've written more code around juju, the gui, the charmstore, etc. I've not written the main juju Go code. I work with those teams a lot.
[18:47] <yosefrow> cool :o
[18:48] <yosefrow> Im confident that Juju / MAAS have a prominent place in the present and future of cloud computing
[18:48] <yosefrow> Without them my life would be a nightmare
[18:52] <rick_h> yosefrow: glad to hear. hopefully the updates get your deploys going smoother
[18:52] <rick_h> https://hangouts.google.com/hangouts/_/kzm7m5vmn5f5lbfm5yr4wkowk4e bdx kwmonroe hml agprado externalreality_  and such for joining the hangout
[18:53] <yosefrow> rick_h, before juju was not on option for certain clients. now it is. I just need to test it
[18:53] <rick_h> https://www.youtube.com/watch?v=Pkbp4VK8-vo for watching in 7 min
[18:53] <rick_h> yosefrow: if you hit anything you know where to find us :)
[18:53] <yosefrow> hehe yep :)
[19:16] <zeestrat> rick_h: A tip on how to build older juju clients would be handy too for us stragglers who need to test things in staging
[19:25] <zeestrat> rick_h: Would love to crib some notes.
[19:25] <rick_h> zeestrat: gotcha, will put together a pastebin for you
[19:25] <zeestrat> Thanks
[19:29] <yosefrow> @rick_h, well that was fun. I originally meant to join as a spectator, but I guess it all worked out. How often do you do this show?
[19:29] <rick_h> yosefrow: every two weeks, same bat time, same bat channel
[19:29] <yosefrow> juju operates on bat time?
[19:29] <rick_h> yosefrow: http://youtube.com/jujucharms check out the playlist for the juju show
[19:29] <rick_h> yosefrow: is there any other? :P
[19:29] <yosefrow> hehe\
[19:32] <rick_h> zeestrat: https://pastebin.canonical.com/209481/ is the gist of what I did
[19:32] <rick_h> zeestrat: that creates a bin directory with a juju in it juju-core_2.2.9/bin/juju
[19:32] <rick_h> I *think* those are the steps needed. I had to monkey around a bit to get it to work but think these are the required ones
[19:36] <yosefrow> rick_h, while you are here, do you know off the top of your head if theres a way to customize the location of the juju gui download automatically downloaded when you bootstrap, or a way to skip the gui being installed so that the installation does not timeout?
[19:36] <yosefrow> this becomes in an issue in ari-gapped environments
[19:36] <zeestrat> rick_h: Getting "You do not currently have access to the pastebin." Internal only?
[19:36] <yosefrow> air-gapped*
[19:36] <rick_h> yosefrow: yes, there's a --no-gui option to bootstrap and then you can always download any gui release tarball and use that as an upgrade-gui command
[19:36] <rick_h> zeestrat: oh doh my bad
[19:36] <rick_h> yosefrow: https://github.com/juju/juju-gui/releases
[19:37] <rick_h> zeestrat: try https://paste.ubuntu.com/26537367/
[19:37] <zeestrat> ty
[19:37] <yosefrow> rick_h, is this the recommended way of performing air gapped installs?, or is a kind of --gui-url option being planned ?
[19:38] <rick_h> yosefrow: no, nothing in the plans. Worth a bug perhaps but unfortunately not a priority there since there's the work around
[19:38] <yosefrow> understood. noting the workaround. Thanks
[19:38] <yosefrow> I wish I joined this channel earlier instead of futzing around on my own for so long. You guys are totally cool
[19:39] <yosefrow> Thanks for all the help
[19:40] <rick_h> yosefrow: no problem, thanks for hacking with stuff and definitely we want people to use our stuff :)
[19:41] <yosefrow> rick_h, This is cutting edge stuff. So I'm glad to be part of the juju movement, even if just as a user/bug reporter.
[19:42] <yosefrow> Ill definitely report bugs and clarify any roadblocks i run into here
[19:42] <rick_h> <3
[20:03] <ryebot> I'm having a hard time getting the cloudinit-userdata model config to work: https://gist.github.com/wwwtyro/77d79e971f8c79590ea47e87b4875d8a
[20:04] <ryebot> I've tried numerous permutations, but I never see anything set in /var/lib/cloud/instance/user-data.txt.i or any of its siblings
[20:05] <ryebot> Any suggestions/examples?
[20:33] <thumper> ryebot: which version of juju are you using?
[20:33] <thumper> hml: ^^
[20:33] <ryebot> @thumper: 2.3.2
[20:33] <ryebot> thumper: I just discovered this: https://github.com/juju/juju/blob/2.3/cloudconfig/userdatacfg_unix.go#L357
[20:34] <ryebot> which invalidates that experiment
[20:34] <ryebot> but I'm trying to get ca-certs to work
[20:34] <ryebot> I'm going to try with final_message next and see if I can even get it to show up
[20:36] <hml> ryebot: packages should be allowed - they are handled separately
[20:36] <hml> ryebot: they are merged into the juju packages list
[20:37] <hml> ryebot: any chance you can check the cloudinit log files on the machine?
[20:37] <ryebot> hml: yeah no problem
[20:38] <ryebot> hml: anything in particular you'd like me to look for?
[20:38] <hml> ryebot: see what references to ack-grep there are.  did it find it in the cloud init files and were there problems installing it
[20:39] <hml> ryebot: you can also see if it make it into /var/lib/cloud/instance/user-data.txt.i
[20:39] <ryebot> hml: no references to ack-grep in /var/log/cloud-init.log or /var/log/cloud-init-output.log
[20:40] <ryebot> hml: it's not in /var/lib/cloud/instance/user-data.txt.i either
[20:40] <hml> ryebot: hrm…
[20:40] <hml> ryebot: trying something in my setup
[20:41] <ryebot> hml sweet thanks
[20:42] <ryebot> hml: fwiw, I wasn't able to get final_message or ca-certs to work either
[20:43] <hml> ryebot: you’re config file worked locally for me
[20:43] <ryebot> well, poop
[20:44] <hml> ryebot: what series are you using?
[20:45] <ryebot> hml xenial
[20:45]  * hml scratching head
[20:45] <ryebot> hml: are you on 2.3.2? or dev branch or?
[20:46] <hml> ryebot: i’m on dev right now - but it works on 2.3.2, went in for 2.3.1, but i’ll double check 2.3.2 in a minute to be paranoid
[20:46] <ryebot> hml: also did you try with a fresh model? I've been reusing the same model and just launching a new machine after each model-config change
[20:47] <hml> ryebot: i had to bootstrap - so it’s all fresh
[20:47] <hml> ryebot: did you bootstrap 2.3.2 or upgrade?
[20:47] <ryebot> gah, my controller is 2.3.1
[20:47] <ryebot> dangit
[20:47] <ryebot> lemme bootstrap a new one
[21:14] <rick_h> ryebot: upgrade! /me did the juju-upgrader tool in the juju show today
[21:14] <rick_h> vs bootstrap, oh except I see there's some question of issues with an upgraded setup
[21:14] <ryebot> rick_h: hmm well I probably should have to add another datapoint, but too late now
[21:15] <ryebot> rick_h: I'll remember next time :)
[22:09] <SuneK> Hi I need some help understanding constraints
[22:09] <SuneK> I was under the impression that when i specify a constraint for an application in a bundle, juju would provision machines that met those constraints?
[22:10] <knobby> that would be my assumption as well. What specific constraint are you using and in what environment?
[22:11] <SuneK> I'm trying to setup the canonical kubernetes distribution
[22:11] <knobby> on bare metal, aws, gce?
[22:11] <SuneK> on vsphere
[22:12] <SuneK> So i loaded the bundle in the juju gui, exported it, and added constraints for the worker nodes, and reimported it to my model
[22:13] <SuneK> But the machines created are way to small
[22:14] <SuneK> I know i could set constraints for the machines themselves, but that's not how it's suppose to be (in my mind at least)
[22:14] <kwmonroe> SuneK!!
[22:14] <knobby> I'm not sure about vsphere and constraints. I thought I saw something about that in an issue recently but I can't find it now. Maybe one of the juju guys can help.
[22:15] <SuneK> kwmonroe: ?
[22:17] <kwmonroe> SuneK: sorry, i forgot to paste the thing i was exlaiming you for...
[22:17] <kwmonroe> so, earlier you asked about this constraint thing, but i wasn't fast enough with the reply before you dropped..
[22:17] <kwmonroe> here's the deal:  bug 1676986
[22:17] <mup> Bug #1676986: juju doesn't honor bundle application constraints (2.1.2) <constraints> <juju:Invalid> <https://launchpad.net/bugs/1676986>
[22:17] <SuneK> Yeah sorry about that; I had to pickup kids :-)
[22:18] <kwmonroe> SuneK: on initial deploy, machine contraints will take prescedent.  application constraints will take effect for stuff like "add-unit"
[22:18] <SuneK> But thanks, I'll look into that
[22:18] <SuneK> Ok, well now there's an explanation, I can work around that
[22:18] <SuneK> Thanks again!
[22:19] <kwmonroe> np