[06:24] <madhukar> Hi All, I am using juju 2.0-beta15. I am trying to deploy a charm bundle. Inside the charm bundle there are few charms which I have pointed to my local charm directory.
[06:25] <madhukar> This is how I have pointed to the local directory:
[06:25] <madhukar>   cassandra:     charm: local:trusty/cassandra
[06:25] <madhukar> and I have updated the $JUJU_REPOSITORY env variable
[06:26] <madhukar> When I deploy this charm, I am getting the below error
[06:26] <madhukar> ERROR cannot deploy bundle: cannot resolve URL "local:trusty/cassandra": unknown schema for charm URL "local:trusty/cassandra"
[06:26] <madhukar> Am I missing something here
[06:26] <madhukar> ?
[06:28] <hloeung> madhukar: what does $JUJU_REPOSITORY point to?
[06:28] <hloeung> and within $JUJU_REPOSITORY, is the charm present as 'trusty/cassandra' (so trusty directory followed by cassandra)?
[06:28] <madhukar> ubuntu@juju-api-client:~$ echo $JUJU_REPOSITORY /home/ubuntu/juju/
[06:29] <madhukar> yeah I have trusty directory followed by cassandra
[06:30] <hloeung> hmm, not sure then. Maybe 'juju --show-log --debug' might show something more helpful
[06:30] <madhukar> ok let me try that
[07:20] <Rajith> hi, In ubunt 14.04, getting error: The requested backend 'zfs' isn't available on your system (missing tools).
[07:23] <Rajith> if I try to install zfsutils-linux, getting error unable to locate package zfsutils-linux.  Let me know how to install zfs file system on ubuntu14.04
[09:14] <junaidali> Hi everyone, I'm deploying a single-controller cluster with juju 2.0 on xenial but percona-cluster seems to have issues. After sometime, other openstack services are not able to connect to mysql. To solve, i have to restart mysql service. There isn't any networking issue as I had deployed single-controller bundle with juju 2.0 on trusty several times on the same setup. Has anyone else facing this issue? or any  idea what might be the cause?
[10:42] <rock> Hi. On juju version, 2.0-rc1-xenial-amd64 I want to deploy a charm bundle. Inside charm bundle several charms are there. I want to point one of the bundle charm to local charm directory. For that How can I exactly edit the bundle.yaml?
[10:47] <zeestrat> rock: Try just using "charm: /path/to/local/charm" instead of "charm: cs:xenial/ceph-mon-3"
[10:48] <rock> zeestrat: Oh. Thank you.
[10:49] <rock> zeestrat: Am I need to change $JUJU_Repository?
[10:51] <rock> I mean am I need to update  $JUJU_REPOSITORY environment variable?
[10:53] <zeestrat> rock: I don't think so if you use an absolute path, but I can't say for sure. P.S. It can be helpful to run juju deploy with the --debug flag to get a bit more info on where Juju is looking for bundle/charms
[10:54] <rock> zeestrat: OK. Thank you.
[11:06] <rock> Hi. I have a question. I have openstack-on-lxd setup with juju version 2.0-beta15. So now I want to enable multipath on nova service deployed LXD container  and cinder service deployed LXD container using our own "cinder-storage" driver charm. But we are facing an issue. http://paste.openstack.org/show/582934/
[11:06] <rock> how to resolve this please?
[11:44] <rock> Hi. # apt-get install multipath-tools --yes   this command is failing on LXD containers. Please anyone tell me the reason for this.
[11:46] <rick_h_> rock: I'm not familiar with the package, but it looks like it might run afoul of the security settings/apparmor bits that allows lxd containers to be isolated cleanly
[11:50] <rock> rick_h_: Thank you for your information. Am I need to change kernal level security settings? But the same package working fine in 16.04 and 14.04 machine.
[11:50] <rick_h_> rock: in a lxd container?
[11:51] <rick_h_> rock: because containers have a shared kernel, there's a lot of work into locking things that could cause issues there.
[11:51] <rick_h_> rock: it might be worth engaging the upstream lxd team on this for better details
[11:52] <rock> rick_h_: Oh. Thank you. How can approach upstream lxd team?
[11:52] <rick_h_> rock: check out https://linuxcontainers.org/
[11:52] <rick_h_> rock: if you go into the lxd xection there's irc, mailing list, etc
[11:54] <rock> rick_h_: Thank you.
[11:54] <zeestrat> rick_h_: Are you guys still aiming for compatibility between RC's so you can upgrade between them?
[11:54] <rick_h_> zeestrat: yes, very much so
[12:01] <zeestrat> rick_h_: Good to hear. Thanks.
[12:57] <lazyPower> cmars: yo lmk when youre ready for another round of refresh, i think we have the ingress lb sorted as well
[13:25] <cmars> lazyPower, cool. i'm having trouble with hacking on local builds of the master & worker charms. system seems to thrash like mad when i upload that 1gb resource :(
[13:25] <lazyPower> well, yeah
[13:25] <cmars> lazyPower, how do y'all develop on them?
[13:25] <lazyPower> wait
[13:25] <lazyPower> 1gb resource?
[13:25] <lazyPower> you're repacking the release tarball right?
[13:26] <cmars> lazyPower, um
[13:26] <cmars> lazyPower, no?
[13:26] <cmars> :)
[13:26] <lazyPower> I sent over a bash script to take that kubernetes releae tarball and split out a worker/master resource package :D
[13:26] <cmars> LOL
[13:26] <cmars> ok
[13:26] <cmars> wow
[13:26] <lazyPower> lol sorry i wasn't too specific on details, something about deadlines, no time, and lack of sleep
[13:26] <cmars> no worries
[13:26] <lazyPower> 1 sec i'll re-gist
[13:26] <cmars> i'm just hackety-hacking on an experiment
[13:26] <lazyPower> https://gist.github.com/fa4a1dca1d313967609bc07183bb272a
[13:27] <cmars> lazyPower, thanks, will take a look
[13:27] <cmars> otoh, i've possibly found a nice way to stress jujud for future profiling >:)
[13:35] <lazyPower> haha sick
[13:35] <lazyPower> cmars: yeah actually we can tank an rc1 container pretty consistently
[13:36] <lazyPower> cmars: around deploy 30/40 of these k8s bundles in succession the controller seems to go unresponsive during a resource upload, and its not clear why
[13:36] <lazyPower> i want to get more time to dig into it and fetch the logs and submit a bug, none to report so far...
[14:43] <geetha> Hi, `juju get-config` command is unrecognized in juju-2.0 rc1.
[14:46] <tvansteenburgh> geetha: it's just `juju config` now (changed in beta18)
[14:50] <geetha> ok, thank you tvansteenburgh..:)
[14:57] <madhukar> Hello
[14:57] <rick_h_> howdy madhukar
[14:57] <madhukar> I am trying to install juju 2.0. However when I try to add the ppa, its giving me the below error
[14:58] <madhukar> ubuntu@juju-api-client:~$ sudo add-apt-repository ppa:juju/devel  Cannot add PPA: 'ppa:~juju/ubuntu/devel'. ERROR: '~juju' user or team does not exist.
[14:58] <rick_h_> there's a firewall issue atm and wonder if that's causing you issues
[14:58] <rick_h_> madhukar: the network is working to get corrected atm
[14:58] <axino> yes this is because of a network outage we're having
[14:58] <axino> we're working on it
[14:58] <rick_h_> ty axino
[14:58] <madhukar> Thanks for the update :)
[14:58] <axino> sorry for the inconvenience
[15:00] <madhukar> No Problem!
[15:02] <madhukar> Any link from where I can track this issue?
[15:02] <axino> madhukar: I'm afraid not
[15:03] <madhukar> hmmm!   Is it possible for somebody to update in this group?
[15:03] <rick_h_> madhukar: there's folks notifying on the launchpad and the juju-gui twitter handles
[15:03] <rick_h_> madhukar: will do
[15:03] <madhukar> ok thanks
[15:22] <madhukar> rick: I m able to add the ppa and install juju 2.0. Thanks for the help!
[15:23] <rick_h_> madhukar: glad it's coming back up
[16:08] <lazyPower> rick_h_: ^5 on getting the topic
[16:11] <stokachu> could probably remove that part now
[16:12] <stokachu> seems it's all back up
[16:12]  * rick_h_ thought he did
[16:12] <stokachu> \o/
[16:17] <lazyPower> cory_fu: i wrote an action in python and need that magic syspath bit that reactive gives me. can you refresh me on what that looks like?
[16:18] <lazyPower> oh i meant updating from beta18 in topic to rc
[16:18] <lazyPower> stokachu: also, hey, we have new kubes
[16:18] <lazyPower> stokachu: wanna give conjure a go and make sure it works as expected?
[16:19] <stokachu> lazyPower: yea ill do a run now
[16:19] <lazyPower> stokachu: ok hang on
[16:19] <lazyPower> we have an edge channel bundle
[16:19] <lazyPower> it fails to resolve elasticsearch when you punch in juju deploy cs:~containers/canonical-kubernetes --channel=edge though
[16:19] <lazyPower> i have no idea why
[16:19] <stokachu> lazyPower: i should update https://github.com/conjure-up/spells/blob/master/observable-kubernetes/metadata.yaml#L5 right?
[16:19] <stokachu> point to canonical one?
[16:19] <lazyPower> or just juju deploy canonical-kubernetes --channel=edge
[16:20] <lazyPower> yep
[16:20] <stokachu> ok
[16:20] <mbruzek> I don't have any idea either, we do have elasticsearch-18 in the store, but I can't get it to deploy out of a bundle.
[16:20] <lazyPower> observable-kubernets has been deprecated in favor of "Canonical distribution of Kubenetes"
[16:20] <lazyPower> stokachu: make sure you use those words, but properly spell kubernetes
[16:20] <lazyPower> it was apparently a big deal last week
[16:21] <stokachu> cool ill get it renamed
[16:25] <stokachu> deploying now
[16:36] <stokachu> lazyPower: mbruzek http://paste.ubuntu.com/23234120/
[16:36] <stokachu> so far so good
[16:37] <mbruzek> stokachu: cool
[16:37] <mbruzek> you attached the resources or is this a charm store deploy?
[16:37] <stokachu> yea this is basically your juju deploy bundle
[16:38] <lazyPower> nice
[16:38] <lazyPower> i think
[16:39] <lazyPower> :D
[16:39] <stokachu> i can do the upgrade-charm with resources if you want
[16:39] <stokachu> to test that
[16:41] <stokachu> looks like i hit an issue with etcd though: http://paste.ubuntu.com/23234120/
[16:41] <stokachu> http://paste.ubuntu.com/23234147/
[16:41] <stokachu> ^ sorry that one
[16:49] <mbruzek> stokachu: That is an old bundle, we have a new one that splits master / worker
[16:49] <lazyPower> stokachu: yeah thats been fixed in the latest etcd. no longer running via rendered shell, i'm using the etcd python api libs
[16:50] <stokachu> https://api.jujucharms.com/charmstore/v5/canonical-kubernetes/archive/bundle.yaml
[16:50] <lazyPower> stokachu: so, this bundle we push this afternoon will land in stable channel, should be nbd then.
[16:50] <stokachu> thats the bundle im using is there another one
[16:50] <stokachu> ah ok
[16:50] <lazyPower> i dont think thats our edge channel bundle
[16:50] <stokachu> yea it's just stable
[16:50] <lazyPower> and the store makes it hard to view bundles in different channels
[16:50] <stokachu> gotcha, yea conjure-up doesn't support edge either
[16:51] <stokachu> ok just ping me here when that bundle is in stable and ill rerun
[16:52] <lazyPower> ok will do stokachu, thanks for TAL so quickly
[16:57] <kwmonroe> hey lazyPower mbruzek, kibana's action is called 'load-dashboard', but i see 'deploy-dashboard' in a couple places (https://jujucharms.com/u/containers/beats-core and chuck's blog: http://insights.ubuntu.com/2016/09/22/monitoring-big-software-stacks-with-the-elastic-stack/).  i don't mind fixing, but how would you like to do it?  symlink deploy->load in kibana, or update the beats-core readme and blog post(s)?
[16:59] <mbruzek> kwmonroe: yeah write a bug for it, so we don't forget
[16:59] <kwmonroe> ack
[17:01] <mbruzek> kwmonroe: symlink would be fine so we don't have to change the blog, but we should also change the readme to be correct.
[17:01] <mbruzek> so both
[17:01] <mbruzek> good find, sorry for the bug
[17:32] <lazyPower> kwmonroe: its a config option these days. probably prefer to load dashboards via config eh?
[17:41] <kwmonroe> hard to say lazyPower, mostly because i don't know kibana very well.  dashboards via config might get dicey if i want beats today and beats plus something else tommorow.  in that case, would i config set dashboards="foo bar beats"?  if i forget "beats" in that string, do those dashboards go away?  if dashboards are meant to be "load once, available forever", then i think an action is appropriate.
[17:41] <kwmonroe> juju my $0.02
[17:42] <kwmonroe> er, *just
[17:42] <lazyPower> kwmonroe: its load once, available forever in both contexts
[17:43] <lazyPower> we supported the config option for bundle deployment to setup the dashboards.. but you bring up an important note
[17:43] <lazyPower> we have introduce immutability into our config
[17:43] <lazyPower> *introduced
[17:43] <kwmonroe> ssssshhhhhhhhhh... mbruzek will hear.
[17:43] <lazyPower> it needs to be said
[17:44] <CorvetteZR1> hi.  i got openstack up and running with openstack-base-xenial-mitaka charm
[17:44] <CorvetteZR1> i can log into the dashboard, but when i go to containers, i get an error:  Unable to get the Swift container listing.
[17:45] <CorvetteZR1> how do i configure this?  how do i log into the servers juju configured?  i can't ssh to them...
[17:51] <lazyPower> CorvetteZR1: juju ssh charm/unit#
[17:51] <lazyPower> eg: juju ssh swift/0
[17:52] <CorvetteZR1> cool, gonna try that
[17:53] <CorvetteZR1> trying to figure out which box it's on :D
[17:53] <CorvetteZR1> ah, it's one of the ceph boxes which failed to deploy...
[17:55] <CorvetteZR1> i can ssh to the node, thanks for the tip lazyPower
[17:55] <CorvetteZR1> not sure if i'll be able to fix my issue from here, but i'll see what i can find :)
[18:01] <lazyPower> CorvetteZR1: happy to help, just let us know, or reach out over the mailing list if nobody seems to be around that knows the answer
[18:01] <lazyPower> protip: juju@lists.ubuntu.com
[18:01] <CorvetteZR1> cool, thanks
[18:02] <lazyPower> tvansteenburgh: standup ping
[18:14] <charles2> nick lazyPower
[20:25] <Siva> I used an yaml file to deploy with juju2.0 and it worked fine
[20:26] <Siva> I am trying the same yaml file with juju1.25 and deployed it using juju-deployer
[20:26] <Siva> I am getting the following error
[20:26] <Siva> ubuntu@juju-1-25-api-client:~$ juju-deployer -c contrail-trusty-3.0.2.1-4-liberty-cs-edit2.yaml -d 2016-09-26 19:58:24 [DEBUG] deployer.cli: Using runtime GoEnvironment on maas 2016-09-26 19:58:24 [ERROR] deployer.cli: Deployment name must be specified. available: ('machines', 'relations', 'series', 'services')
[20:26] <Siva> Any idea what is wrong here?
[20:27] <Siva> Any help is much appreciated
[20:28] <rick_h_> Siva: last year we updated the bundle format to v4 and juju 2.0 uses that. See https://blog.jujugui.org/2015/08/13/bundles-bundles-bundles/
[20:29] <Siva> does it mean juju-deployer does not support format v4?
[20:31] <rick_h_> Siva: it does support the v4 format
[20:31] <rick_h_> Siva: sorry, I read that backwards
[20:32] <rick_h_> Siva: that you had an old bundle that juju 2.0 would not accept
[20:32] <rick_h_> Siva: what version of 1.25? and maybe you need to s/applications/services in the bundle to get it to work on 1.25?
[20:34] <Siva> This is my juju version
[20:34] <Siva> ubuntu@juju-1-25-api-client:~$ juju --version 1.25.6-trusty-amd64
[20:34] <rick_h_> Siva: right, what does the bundle look like?
[20:34] <rick_h_> Siva: services were renamed applications in 2.0 so if it's "applications" I don't know if the deploy will remap that for you
[20:37] <Siva> It looks like this
[20:37] <Siva> http://pastebin.ubuntu.com/23235034/
[20:37] <Siva> This is not the full yaml file but something like this
[20:37] <Siva> This is not the full yaml file but something like the above
[20:40] <Siva> rick_h_: Are u able to look at the yaml file pasted in pastebin?
[20:40] <Siva> What should I change there to make it work with Juju 1.25
[20:45] <lazyPower> stokachu: ping re: kubernetes being ready for a go
[20:47] <Siva> rick_h_: Do you want me to paste the full yaml file? Please let me know
[20:54] <rick_h_> Siva: hmm, ok yea that looks about right
[20:54] <rick_h_> Siva: not sure on the series outside
[20:54] <rick_h_> Siva: that might be what's confusing it
[20:56] <Siva> Do you mean the 'series:trusty
[20:56] <Siva> Do you mean the 'series:trusty' in the very first line?
[20:57] <rick_h_> Siva: yes, try without that?
[20:58] <Siva> OK. Tried. I am still getting the same error
[21:01] <rick_h_> Siva: hmm, ok. so the error from the deployer is fussing about the keys
[21:02] <Siva> What does that mean?
[21:04] <rick_h_> Siva: sorry, on the phone. I mean to double check the keys are named matching the deployer error
[21:08] <Siva> @rick_h_, yes all the keys are available in yaml file
[21:08] <Siva> so don't know why it is complaining that it must be specified
[21:21] <kwmonroe> Siva: are you running this?   juju-deployer -c contrail-trusty-3.0.2.1-4-liberty-cs-edit2.yaml -d
[21:31] <kwmonroe> Siva: was able to able to start the deploy with your pastebin'd yaml.  looked ok to me (it eventually failed on undefined relations, but i attribute that to not being your full yaml file).  anyway, i'm running juju-1.25.6-trusty-amd64 and  juju-deployer-0.9.0~bzr193~58~ubuntu14.04.1
[21:41] <Siva> kwmonore:  yes I ran the command 'juju-deployer -c contrail-trusty-3.0.2.1-4-liberty-cs-edit2.yaml -d'
[21:42] <Siva> I am using the following version of juju-deployer
[21:42] <Siva> ubuntu@juju-1-25-api-client:~/.juju$ dpkg -l | grep deployer ii  juju-deployer                    0.6.4~bzr168~49~ubuntu14.04.1    amd64        A tool to deploy complex stacks of services using juju
[21:42] <Siva> 0.6.4~bzr168~49~ubuntu14.04.1
[21:43] <Siva> I am using the same version of juju 1.25 you are using as well
[21:44] <Siva> How do I install the version of juju-deployer you are using?
[21:48] <kwmonroe> tvansteenburgh: Siva's asking about juju-deployer v0.9.  i got that from your ppa.  is that the best place to grab it?
[21:48] <kwmonroe> ...backscroll for context
[21:50] <kwmonroe> Siva: unless tvansteenburgh says different, you can get 0.9 with:  add-apt-repository -y ppa:tvansteenburgh/ppa
[21:52] <Siva> OK. I will try with the version you are using an see if it works
[21:52] <Siva> I feel may be the version of juju-deployer I am using is having some issues
[21:54] <kwmonroe> yeah Siva, the error message you got ("Deployment name must be specified"), sounds an awful lot like juju-deployer isn't recognizing your v4 bundle format.  not sure when that support was introduced, but i know that juju-deployer-0.9 definitely supports v4.
[22:10] <Siva> kwmonore: I destroyed my existing container and doing it fresh. Give me 10 more mts. I will update you
[22:10] <Siva> sorry for the delay
[22:11] <veebers> alexisb: Do you have the link for the bug filed by OIL you mentioned in the stand up? (race condition with adding/removing/listing models)
[22:15] <alexisb> veebers, yes
[22:18] <alexisb> veebers, https://bugs.launchpad.net/juju/+bug/1618212
[22:18] <mup> Bug #1618212: juju models fails during model destruction <oil> <oil-2.0> <juju:Triaged> <https://launchpad.net/bugs/1618212>
[22:18] <alexisb> thumper, I will jump back on the call
[22:19] <veebers> alexisb: sweet thanks, I think this is the bug I see as well
[22:27] <Siva> kwmonroe: Thank you so much. I am not getting that error anymore
[22:27] <Siva> It is deploying it now
[22:27] <Siva> so the version of juju-deployer you pointed out has support for the v4 bundle format
[22:48] <kwmonroe> glad it's working Siva!
[22:49] <magicalt1out> liar