[00:59] is there anyway to change the admin_domain for keystone ? it seems that keystone charm creates admin_domain as default. I need to have this charm create the domain "default" instead of admin_domain === scuttlemonkey is now known as scuttle|afk [07:46] Good morning Juju world! [08:15] morning all. i have deployed an openstack cloud via juju. now i want to deploy services adding the same openstack in the juju cloud [08:15] unfortunately it fails [08:16] juju add-cloud openstack1 tenant-demo-env.yaml ERROR cannot unmarshal yaml cloud metadata: yaml: unmarshal errors: line 4: cannot unmarshal !!str `userpass` into []cloud.AuthType [08:16] this is the yaml [08:16] https://paste.ubuntu.com/24724275/ === Salty is now known as Guest45701 [09:00] ok, i fixed it by using [userpass] syntax [09:00] Hi, Whenever I create an new service, the first TYPICAL ERROR message I get from juju debug-log reads like follows: "unit-name-1: 11:54:06 ERROR juju.worker.dependency "metric-collect" manifold worker returned unexpected error: failed to read charm from: /var/lib/juju/agents/unit-name-1/charm: stat /var/lib/juju/agents/unit-name-1/charm: no such file or directory" === disposable3 is now known as disposable2 === mimizone_ is now known as mimizone === SaMnCo_ is now known as SaMnCo [09:01] However, Sshing to the related machine, and checking the full, it's found indeed however, the owner and the group are ROOT:ROOT [09:01] root@juju-8aa58b-21:/var/lib/juju/agents/unit-name-1/charm# pwd /var/lib/juju/agents/unit-name-1/charm [09:02] root@juju-8aa58b-21:/var/lib/juju/agents/unit-name-1/charm# ll total 583 drwxr-xr-x 15 root root 30 May 31 08:54 ./ drwxr-xr-x 4 root root 7 May 31 08:54 ../ drwxr-xr-x 2 root root 8 May 31 08:54 actions/ -rw-r--r-- 1 root root 670 May 31 08:54 actions.yaml drwxr-xr-x 2 root root 3 May 31 08:54 bin/ -rw-r--r-- 1 root root 249687 May 31 08:54 name.svg -rw-r--r-- 1 root root 6659 May 31 08:54 config.yaml -rw-r--r [09:05] Is it because, no metrics are yet defined on such charm? But why would it necessitates to call metric-collect as first thing? [09:15] Naz, that looks like this bug: https://bugs.launchpad.net/juju/+bug/1656258 [09:15] Bug #1656258: metric-collect returned unexpected error

[09:21] @Mattyw, Yes, correct, thank you, it's mentioned that this bug is fixed in 2.2.rc1 and won't be fixed in 2.1.2 [09:21] Naz, I don't think it's a problem though - the unit should continue working as expected [09:21] @Mattyw, Yup, correct :) [09:23] Excuse me, I am still learning, I shall check the issues in bug repositories next time :) [09:23] Naz, no problem - I only knew the bug was there because I remember discussing it a few weeks ago [09:24] @Mattyw, Great, thank you my friend, you are the man :) [09:26] I see: "The current stable version of Juju is 2.1.3 and is recommended for everyday production use." [09:26] I am running on 2.1.2! How could I upgrade to latest stable without losing my models/apps? [09:27] https://jujucharms.com/docs/stable/reference-install, isn't it for install from scratch? [09:36] Found it :), juju upgrade-juju --agent-version 2.1.3 [09:51] Question regarding network spaces: afaict juju/state/machine_linklayerdevices.go , it seems that in order for a space to be considered as available on a host, that host must have a iface *with an ipaddr* in the spaces subnet, is that correct? [09:55] The problem is, I'd like to have a metal deployed by maas 2.2, with a bridge iface in a vlan'ed space but w/o ipaddr, and handing out ipaddr only in containers [09:55] Those are pub ipv4 addr and therefore scarce [09:57] Getting an error message when deploying a test container -- cannot start instance for machine "1/lxd/2": unable to setup network: host machine "1" has no available device in space(s) "space-api" [09:59] But, icb reading that source wrong, IANAGoDeveloper [10:00] Fwiw, this is maas 2.2 and juju 2.1.3 === MmikeM is now known as Mmike [10:43] Interestingly, in the above ^^ ctxt, the constraints matcher seems less picky. I can add-machine --constraints spaces=space-api and get the machine deployed (it warns about an address-less device though) [11:15] chrome0: so I know that the goal is to support what you're doing. I don't think we're there to support it yet. jam is the best person to confirm if there's any trick to getting it working under current conditions [11:15] maybe wpk [11:21] rick_h: I'm missing the context from chrome0 had to restart my comp [11:21] jam: ah sorry, he's wanting to use maas to get the host w/o an ipaddr on the device passed to the container [11:23] rick_h: chrome0: wpk has been digging more closely into being able to represent what space a device is in when it doesn't have an IP address, but right now, when we try to link up charms to devices, etc [11:23] we explicitly enumerate devices, find their IP addresses, and then compare them to the subnets we know about [11:23] we want to change that, but its the current logic [11:23] so again, you can probably get the right machine [11:24] but things like binding the charm to the right device won't quite work right [11:29] ty jam [11:54] axw_ develop branch seems in a high activity right now... [11:54] axw_ Updates were rejected because the tip of your current branch is behind its remote counterpart. [11:57] axw_ solved [12:25] Thanks rick_h and jam , was afraid you'd say that. Is there a bug filed about this? If not I'll create one [12:47] chrome0: there is a bug, but i'd have to look it up [12:48] Cheers, I'll see if I can find it [12:49] chrome0: https://bugs.launchpad.net/juju/+bug/1659376 [12:49] Bug #1659376: should be a way to specify space binding for maas "unconfigured" interface

[12:49] I think is the one I was thinking [12:52] Yup, my usecase is a bit different but this should cover it === scuttle|afk is now known as scuttlemonkey [14:50] SimonKLB: i summon thee (if you're summonable) [14:50] lazyPower: whats up? [14:51] SimonKLB: i'm tapping Cynerva to lend a hand with our networking conundrum while i step away to run some quick errands [14:51] o/ [14:51] hey! [14:51] Cynerva: this is simon, he's got an issue with pod networking in lxd [14:51] howdy [14:51] okay [14:51] i thought it was kube-proxy related but that doesn't appear to be the case [14:51] cant resolve VIP's in a lxd worker's pods :| [14:51] hmm [14:52] i am able to access the service IPs from the worker host [14:52] so i think the forwarding is fine [14:52] service -> pod that is [14:53] but pod -> service is not working [14:53] okay, thinking for a bit [14:53] also, i can see the traffic on the cni0 interface [14:54] i'm trying to remember, i've seen something like this before but don't remember the circumstances [14:55] is this across workers? or pod->service within a single worker? [14:55] single worker [14:59] okay, i dunno [14:59] i'm suspicious kube-proxy is still doing something weird [14:59] with traffic that comes from the pod IP range [15:00] SimonKLB: can you send me the output of `iptables-save` and `cat /var/snap/kube-proxy/current/args` from the worker? [15:00] yup [15:44] hello jujuers. I have a couple of subordinate charms in "allocating - waiting for machine" state (weird for a subordinate! the machine it's being deployed to has already been provisioned and the main service is ok). The log says this: 2017-05-31 08:29:17 ERROR juju.worker.proxyupdater proxyupdater.go:164 can't connect to the local LXD server: LXD socket not found; is LXD installed & running? [15:44] any ideas why it seems to want lxd? [16:12] hi all, i had to open a bug for kubernetes bundle not honoring proxy settings [16:12] https://bugs.launchpad.net/bugs/1694743 [16:12] Bug #1694743: kubernetes-[master,worker] snap install fail due to missing proxy [16:31] o/ juju world [18:07] Hey quick question if anyone has the time... I'm just running the base openstack bundle on juju and wondering how I connect to the cloud it deploys? I can figure out how to shell into the nodes, but have no idea where the credentials are for the openstack cloud [18:15] hi vlad_ - there are notes in the Readme for the openstack-base bundle about how to connect. See the section "Accessing the cloud" at https://jujucharms.com/openstack-base/ [18:16] actually, start two paragraphs earlier at "Ensuring it's working" :) [18:17] it suggests you download the bundle, but you could alternatively just get this file https://api.jujucharms.com/charmstore/v5/openstack-base/archive/novarc [18:17] for the default env vars that'll let the clients log in [18:18] Ahhh that's the exact file I was looking for sorry didn't see the link to it my mistake [18:19] mmcc: thanks kindly [18:19] vlad_: you're welcome, glad I could help! === Guest____ is now known as bladernr-lex [21:00] I realize that this probably goes against the cattle vs pet convention, but is it possible to specify the hostname or naming convention for a new machine?