[08:14] <kjackal> Good morning Juju world!
[08:16] <hoenir> good morning to you kjackal
[09:26] <kklimonda> is there a way to define, and later reference, variables in a bundle file?
[09:27] <kklimonda> it's painful to embed gpg key for each and every service
[09:37] <kjackal> kklimonda: I haven't seen variables in a bundle
[09:52] <cnf> morning
[10:17] <kjackal> hi cnf
[10:20] <cnf> another week, another try at getting stuff to work :P
[10:56] <cnf> is there any documentation on how to put the juju api behind a reverse proxy?
[11:15] <alexlist> Hi... just trying Juju-as-a-Service. At the deploy step, 10mins ago I got an option to choose the target public cloud. Now I have this spinning circle and don't get any options to choose. Are we being slashdotted? ;)
[11:20] <alexlist> I see nothing on Twitter regarding JaaS ...
[11:34] <rick_h> alexlist: not that I'm aware of. Are there any errors in the JavaScript console? Did you get.to the point to enter credentials?
[12:30] <zeestrat> kklimonda, kjackal: We've been using variables for a short while now after coming over this example bundle: https://launchpadlibrarian.net/298175262/bundle.yaml
[13:10] <alexlist> rick_h: I get to the point "Choose cloud to deploy to" but no buttons to choose gce,aws,azure anymore. That worked earlier today. http://pastebin.ubuntu.com/24170595/
[13:11] <rick_h> alexlist: and this is on jujucharms.com/something ?
[13:12] <rick_h> alexlist: can you please post a bug at https://github.com/CanonicalLtd/jujucharms.com
[13:12] <alexlist> rick_h: will do
[13:12] <rick_h> alexlist: screenshots, urls, etc that would be really helpful
[13:35] <kklimonda> zeestrat: thanks, that does seem to work :)
[13:40] <zeestrat> kklimonda: np :) You reminded me that I had should add a bug for docs on that. See https://github.com/CanonicalLtd/jujucharms.com/issues/433
[13:41] <zeestrat> kklimonda: Came over that example bundle by luck in the #openstack-charms channel I think.
[13:54] <kjackal> nice zeestrat I did not know you could do that! Cool!
[14:00] <alexlist> rick_h: https://github.com/CanonicalLtd/jujucharms.com/issues/434
[14:01] <alexlist> rick_h: I got this far using Firefox. Will restart my Chrome to see if it's something in the cache that stopped me from selecting the cloud provider.
[14:02] <zeestrat> kklimonda, kjackal: Somewhat related: Passing dynamic stuff (perhaps --config) at deploy time for bundles is on the roadmap according to Rick.
[14:27] <jamespage> rick_h: hey - aware of any v4 charmstore API type issues?
[14:27] <jamespage> we just started hitting
[14:27] <jamespage> https://www.irccloud.com/pastebin/gO5ykMTA/
[14:27] <rick_h> jamespage: not yet, looking
[14:27] <beisner> hi all.  Is https://api.jujucharms.com/ --> 'Not Found' a known issue?
[14:28] <beisner> oh /me sees james just asked here.  thx
[14:29] <rick_h> beisner: yea, looking. https://api.jujucharms.com/charmstore/v4/~openstack-charmers-next/percona-cluster/meta/any loaded for me
[14:29] <rick_h> is that loading for you beisner jamespage ?
[14:29] <jamespage> rick_h: yup
[14:29] <rick_h> hmm, putting series back in worked: https://api.jujucharms.com/charmstore/v4/~openstack-charmers-next/xenial/percona-cluster/meta/any
[14:29] <rick_h> I wonder if something flipped...
[14:29] <beisner> rick_h, yes, but this is not:
[14:29] <beisner> https://api.jujucharms.com/v4/~openstack-charmers-next/xenial/percona-cluster/meta/any
[14:29] <jamespage> s/charmstore//g
[14:30] <beisner> which is the url we get when using bundletester + amulet + theblues + charmstore et al
[14:30] <rick_h> beisner: oic, with the /charmstore
[14:31] <rick_h> beisner: jamespage ok, so the /charmstore is supposed to be there. It says what service API to hit. I'm not sure if that changed recently but it kind of worked w/o by accident due to "first thing in the list wins" rules.
[14:31] <rick_h> beisner: jamespage I'll see if something's changed and if we can get the "magic" behavior back, but we need to file bugs/get tools updated to use the right urls.
[14:38] <jamespage> rick_h: ack
[14:39] <jamespage> rick_h: please on restoring magic
[14:39] <jamespage> rick_h: gate blocked atm
[14:39] <rick_h> jamespage: working on a cowboy to production to add a rewrite rule atm
[14:39] <rick_h> jamespage: will try to get it back asap with webops help
[14:44] <beisner> jamespage, as a work-around, i can have osci set env var for libcharmstore to consume @ https://github.com/juju-solutions/libcharmstore/blob/master/charmstore/lib.py#L37
[14:45] <beisner> that should unblock on the short term, but we should defo raise an issue @ libcharmstore (i think) to track the real fix
[14:48] <rick_h> beisner: ty for finding the bad url source. I'll get a PR to them.
[14:48] <rick_h> RT is filed for the apache redirect, will hopefully get that through soon
[14:52] <beisner> rick_h, jamespage - raised https://github.com/juju-solutions/libcharmstore/issues/5 -- will have a PR up momentarily to jig that default global URL.
[14:54] <rick_h> beisner: so there's a default url in theblues that the libcharmstore should just reuse
[14:54] <rick_h> beisner: was going to do a pr for it once I get the RT picked up
[14:54] <beisner> rick_h, cool, i'll hold off, let you drive that.  thx
[15:09] <cnf> anyone know how to run the juju api behind a reverse proxy?
[15:32] <rick_h> beisner: jamespage please check it again, cowboy in place
[15:35] <beisner> rick_h, yeehaw!  and yep, the url works now.  thx
[16:03] <mbruzek> bdx: is it too early to ping?
[16:52] <hoenir> juju can add storage to machine with add-storage?
[16:53] <hoenir> or this options is only with units that support..
[16:53] <hoenir> this option?
[17:03] <stormmore> o/ juju world
[17:51] <bdx> mbruzek: sup
[17:52] <mbruzek> bdx: I heard you have your own haproxy charm? Is that right?
[17:54] <bdx> mbruzek: I wish I was that cool
[17:54] <mbruzek> bdx: I am working on reworking the Canonical Distribution of Kubernetes to use haproxy load balancer instead of nginx. And my experience with haproxy is minimal. Wondering if you had any tips/tricks or advice/pointers.
[17:55] <bdx> mbruzek: https://github.com/jamesbeedy/juju-layer-lxd-proxy
[17:55] <bdx> ^ non haproxy based
[17:55] <bdx> just uses iptables, but that may be what you are referencing
[17:56] <mbruzek> bdx:  I was misinformed. This charm looks super interesting in its own right.
[17:56] <bdx> mbruzek: when I was setting up deis, I had to manually configure haproxy to proxy to the deis router/component ports on each of the nodes
[17:57] <mbruzek> oh maybe that was the intel I got. OK
[17:57] <bdx> mbruzek: let me see if I can find that for you for reference real quick
[17:57] <jrwren> mbruzek: haproxy charm is very nice. I got a bit familiar with it. I cna help too if needed.
[18:00] <cholcombe> can juju deploy centos7 at this point?  i thought it could in the past but i can't seem to find the centos image
[18:04] <mbruzek> jrwren thanks I might hit you up.
[18:04] <mbruzek> bdx: I would be interested in the configuration if you could share it.
[18:05] <bdx> mbruzek: here is an example with multiple frontends and backends ... this (multiple frontend/backend) can be accomplished with the haproxy charm too -> http://paste.ubuntu.com/24171930/
[18:07] <tvansteenburgh> mbruzek: here's the haproxy config from the revq bundle, in case that helps http://pastebin.ubuntu.com/24171951/
[18:25] <bdx> mbruzek: I'm currently working on some haproxy relations for one of my apps to accomplish creating the multiple frontend -> multiple backend routing via juju relations, I'll keep you posted when I have something worth sharing
[18:26] <mbruzek> bdx that would be great. Thanks
[19:19] <skayskay> is there a way to remove a resource?
[19:19] <jrwren> no. there is no way.
[19:20] <skayskay> ouch. so, if I have a charm that uses the snap layer, and I've attached a snap so that I can test the snap before publishing it, then I am stuck using a local file after that and can't go back to using the snap from the store
[19:21] <skayskay> I want to be able to switch
[20:38] <skayskay> I'm trying to remove-application and it's not working
[20:39] <skayskay> I can resort to removing the machine by force, but I should figure out why the application won't go away
[21:14] <zeestrat> skayskay: Could you try pasting "juju status name-of-your-application --format yaml"?
[21:15] <skay> zeestrat: it's not in the same state. I gave up and removed the machine
[21:15] <zeestrat> skay: Gotcha. I've run into such situations where it was stuck upgrading.
[21:16] <skay> zeestrat: aha, my upgrade step was failing
[21:16] <skay> zeestrat: I released a new charm to pick up a fix in the snap layer
[21:17] <skay> zeestrat: what did you end up doing?
[21:19] <zeestrat> I got hit by #1671428 so I ended up killing the controller. I logged #1671476 for being stuck while upgrading.
[21:19] <mup> Bug #1671428: deploying with default binding prevents upgrade-charm <juju:Fix Committed by jameinel> <juju 2.1:Fix Committed by jameinel> <https://launchpad.net/bugs/1671428>
[21:19] <mup> Bug #1671476: destroy-model does not work when charm is stuck in upgrading state <juju:New> <https://launchpad.net/bugs/1671476>
[21:20] <skay> zeestrat: did you try remove-machine with --force? that seemed to work for me
[21:21]  * skay adds self to bug
[21:21] <zeestrat> skay: Not yet. I can give it a go.
[21:23] <kklimonda> ha, I was just about to report that bug
[21:43] <Guest20118> I am rendering a list of ipaddress in my charm template
[21:44] <Guest20118> I want the rendered output to be like this servers = ["192.168.1.252", "192.168.1.253", "192.168.1.254"]
[21:44] <Guest20118> I have my template entry like this 'servers = [{{ '\"'~servers|join('\", \"')~ '\"' }}]
[21:45] <Guest20118> My input.yaml file to 'juju config <application> --file= input.yaml' looks like this
[21:46] <Guest20118> servers: "192.168.1.252 192.168.1.253 192.168.1.254"
[21:46] <Guest20118> The rendered output looks like this
[21:46] <Guest20118> servers = ["1", "9", "2", ".", "1", "6", "8", ".", "1", ".", "2", "5", "2", " ", "1", "9", "2", ".", "1", "6", "8", ".", "1", ".", "2", "5", "3", " ", "1", "9", "2", ".", "1", "6", "8", ".", "1", ".", "2", "5", "4"]
[21:46] <Guest20118> in the config.yaml it is of type string
[21:48] <Guest20118> In the charm code i, yaml load it like this servers_list = yaml.load(config.get("servers"))
[21:48] <Guest20118> What is wrong here?
[21:48] <Guest20118> Why is very character interpreted a string here?
[21:49] <Guest20118> Any help is much appreciated
[22:11] <tvansteenburgh> Guest20118: servers = {{ servers }}
[22:11] <tvansteenburgh> it's already a list
[22:18] <tvansteenburgh> Anyone know how to find out in which regions a particular aws instance type is available?
[22:20] <tvansteenburgh> I'm trying to deploy a p2.xlarge in us-east-1 (juju 2.1.0) controller and I'm getting 'invalid constraint value: instance-type=p2.xlarge'
[22:26] <tvansteenburgh> Well I just deployed a p2.xlarge into us-east-1 with the AWS console, so I'm not sure why Juju won't do it. :/
[22:27] <anastasiamac> tvansteenburgh: instance types support was updated in juju 2.1.1
[22:28] <tvansteenburgh> anastasiamac: Oooh. I was trying to remember if they were baked in or downloaded from somewhere - I guess it's the former.
[22:28] <tvansteenburgh> anastasiamac: Thanks!
[22:28] <anastasiamac> tvansteenburgh: it requires manual intervention. but we try to keep on top of it for every new release
[22:29] <tvansteenburgh> anastasiamac: So there's no way for me to use a p2 instance type w/o upgrading juju?
[22:29] <anastasiamac> tvansteenburgh: potentially
[22:39] <tvansteenburgh> anastasiamac: I upgraded to 2.1.1 and get the same error. Any ideas?
[22:39] <tvansteenburgh> doh
[22:44] <tvansteenburgh> anastasiamac: I upgraded to 2.1.1 and get the same error. Any ideas?
[22:45] <anastasiamac> tvansteenburgh: wallyworld might have a better idea if p2.xlarge was added .. m sorry i have issues with my machine atm, can't look at codebase
[22:46] <anastasiamac> tvansteenburgh: i can confirm a bit later on
[22:46] <tvansteenburgh> anastasiamac: k thanks