[01:01] <[Kid]> rick_h, you around/ === wgrant__ is now known as wgrant === frankban|afk is now known as frankban [08:51] good moring [08:52] I have a problem with juju after upgrading to 2.2.5 [08:52] got "no matching agent binaries available" when trying to deploy a model [08:54] destroyed all controllers and tried to reinstall juju and then do a bootstrap, but same problem [08:54] tamas@tamas-EasyNote-TE11HC ~ $ sudo snap install juju --classic [08:54] [sudo] password for tamas: [08:54] juju 2.2.5 from 'canonical' installed [08:54] tamas@tamas-EasyNote-TE11HC ~ $ sudo snap list juju [08:54] Name Version Rev Developer Notes [08:54] juju 2.2.5 2713 canonical classic [08:54] tamas@tamas-EasyNote-TE11HC ~ $ juju bootstrap [08:54] Clouds [08:54] aws [08:54] aws-china [08:54] aws-gov [08:54] azure [08:54] azure-china [08:54] cloudsigma [08:55] google [08:55] joyent [08:55] localhost [08:55] oracle [08:55] rackspace [08:55] Select a cloud [localhost]: [08:55] Enter a name for the Controller [localhost-localhost]: localhost [08:55] Creating Juju controller "localhost" on localhost/localhost [08:55] Looking for packaged Juju agent version 2.2.5 for amd64 [08:55] To configure your system to better support LXD containers, please see: https://github.com/lxc/lxd/blob/master/doc/production-setup.md [08:55] ERROR failed to bootstrap model: no matching agent binaries available [08:55] tamas@tamas-EasyNote-TE11HC ~ $ cat /etc/issue [08:55] Linux Mint 18 Sarah \n \l [10:07] Hi All, Is there a way to tag charms on charmstore for specific release for e.g. tag the uploaded charm for OCATA release only. [10:59] Hello, does anyone know if the maas-image-builder is still available? [10:59] bzr branch lp:maas-image-builder [10:59] bzr: ERROR: Invalid url supplied to transport: "bzr+ssh://bazaar.launchpad.net/+branch/maas-image-builder": no supported schemes [11:18] Actually the MaaS docs say I need Ubuntu Advantage for custom images. [11:54] Any of you seen the message: ERROR failed to bootstrap model: no matching agent binaries available [11:54] after upgrading to 1.2.5 ? [11:54] sry I mean 2.2.5 [13:01] <[Kid]> is there a way to deploy a juju HA cluster with MAAS to a lxd container? [13:21] hi - did anyone run into this before (and has an idea about it): [13:21] "juju status" [13:21] "ERROR unable to connect to API: Forbidden" [13:21] ssh login to the controller and 'juju controllers' still work, but more or less no other juju cmd anymore ... [14:32] anyone want to hazard a guess as to why trying to bootstrap k8s on lxd via conjure up says [14:32] Error bootstrapping controller: ['ERROR enabling HTTPS listener: cannot listen on https socket: listen tcp [::]:8443: bind: address already in use'] [14:32] but [14:32] fuser 8443/tcp [14:32] is empty [14:34] magicaltrout: do you have multiple lxd's installed, like deb and snap? [14:34] conjure-up requires snap lxd [14:35] good catch [14:35] thanks stokachu [14:35] magicaltrout: when snap lxd is first class in next lts this wont be a problem [15:03] <[Kid]> is there a way to put SSH keys into a credentials.yaml for my own private cloud provider? [15:04] [kid]: use `juju add-ssh-key "$(cat mykey.pub)"` [15:05] that will add the key to your model [15:05] <[Kid]> bdx, awesome, i think that's what i needed [15:05] <[Kid]> bdx, will this allow me to bootstrap juju onto another machine? [15:06] <[Kid]> that has that key [15:06] oh [15:06] I can't be sure [15:06] if its the default ssh of of your user [15:06] in ~/.ssh/id_rsa.pub [15:06] I think it gets added automatically [15:06] don't quote me on that [15:07] :) [15:07] <[Kid]> yeah, i have two machines and i am using one to bootstrap the other, then i will add the other one to the model and join them in a HA cluster [15:07] <[Kid]> well, when i do bootstrap, it is trying to use the key at /home/ubuntu/.local/share/juju/ssh/juju_id_rsa.pub [15:08] [kid]: again, don't quote me on this, but I *think* you need a quorum of at least 3 for controller HA [15:08] <[Kid]> ohh dammit [15:08] ahh I see [15:08] yeah, so add /home/ubuntu/.local/share/juju/ssh/juju_id_rsa to you ssh keys [15:08] then you will just be able to ssh [15:09] <[Kid]> well, that's what i was thinking. [15:09] eval "$(ssh-agent -s)" [15:09] <[Kid]> i have to do that on the remote host [15:09] $ ssh-add -K /home/ubuntu/.local/share/juju/ssh/juju_id_rsa [15:12] <[Kid]> that just adds the juju client key for each local host [15:13] <[Kid]> i need to add the keys from node 1 to node 2 and vice versa [15:14] <[Kid]> that worked. [15:14] <[Kid]> i just did it manually [15:15] <[Kid]> the question now, is there a way to have juju use the same client key? [15:15] <[Kid]> on all of its deployments [15:21] yeah [15:21] add it to the model [15:32] <[Kid]> oh with the command you said earlier? [15:44] [Kid]: yea [16:08] goodevening from sweden [16:08] Any of you seen the message: ERROR failed to bootstrap model: no matching agent binaries available [16:08] after upgrading to version 2.2.5 [16:09] tamszagot: known issue related to the bionic update [16:09] ahh , thanks [16:10] is it possible to downgrade? [16:10] to precious version? [16:10] previous [16:11] tamszagot: the trick is to specify the series as xenial at the moment. You can pass --series into the deploy command or the add-machine command or use the charm urls with series in them such as cs:xenial/postgresql [16:12] tamszagot: the team's actively getting new releases out that corrects this [16:12] thank you very much rick! [16:14] hmm ... I have no controller since I uninstalled juju [16:14] and --series doesn't seme to be an option of the bootstrap command [16:15] (I'm new with juju so bare with me ..) [16:15] juju bootstrap --bootstrap-series xenial [16:15] tamszagot: ^^ [16:15] thanks [16:16] (I would have changed that --bootstrap-series to plain --series if you asked me :) [16:22] tamas@tamas-EasyNote-TE11HC ~ $ juju controllers [16:22] Use --refresh flag with this command to see the latest information. [16:22] Controller Model User Access Cloud/Region Models Machines HA Version [16:22] localhost* default admin superuser localhost/localhost 2 1 none 2.2.5 [16:22] gQuigs: thanks alot! [16:23] :) === frankban is now known as frankban|afk [17:37] 20 minutes until the Juju show, get your youtube ready https://www.youtube.com/watch?v=zhz1PSXyIuA [17:51] https://hangouts.google.com/hangouts/_/laeoptm2grfy5lpnc67r3yld3me to join the conversation [17:51] starting in 9 [17:51] wpk: bdx kwmonroe and others ^ [17:52] i'll be there rick_h -- just gotta find some pants. [17:53] kwmonroe: oh good idea :) [18:00] bdx: you on the way? [18:02] <[Kid]> don't need pants, just point the camera on the upper half [18:45] does anyone know of a good example of using Juju storage with the MAAS provider? [18:45] I found this https://jujucharms.com/docs/devel/charms-storage#maas-(maas) [18:46] but it doesn't really give any examples, I'm going to be digging into it a bit [18:47] just wondering if anyone has made any headway there that I might be able to take a look at === atrius_ is now known as atrius [19:58] good afternoon [19:59] I'm trying to do 'juju bootstrap lxd lxd-local' but it gets stuck in waiting for address 20m, after which it gives 'failed to bootstrap model: without getting any addresses' [19:59] kubuntu 17.04 host [20:00] juju --version: 2.2.5-zesty-amd64 [20:00] lxd and lxc --version: 2.19 [20:01] lxc profile list shows two profiles: default and juju-controller [20:02] lxc network list shows lxdbr0, as type BRIDGE, managed [20:13] When creating a GCE instance Is there a way to set up a new instance to run as a service account? [20:25] halp [20:26] charlls: morning [20:26] aemara: what do you mean by service account? [20:33] hi thumper [20:33] I'm trying to do 'juju bootstrap lxd lxd-local' but it gets stuck in waiting for address 20m, after which it gives 'failed to bootstrap model: without getting any addresses' [20:34] kubuntu 17.04 host, juju --version: 2.2.5-zesty-amd64, lxd and lxc --version: 2.19 [20:35] ok... [20:35] likely to be one of several problems [20:35] however [20:35] firstly [20:35] if you try the bootstrap in one window, [20:35] and then in another terminal [20:35] do 'lxc list' [20:35] do you see the machine? [20:35] @thumper basically its is a way to authenticate and authorize the instance to use other GCE services without having to login. I am trying to use the kubernetes charm and I have a private docker registry setup on GCE. To access it the instance need to have a service account which juju does not create or attach to default. https://cloud.google.com/compute/docs/access/service-accounts [20:36] if you do, you want to get into that container and check the cloud init output [20:36] lxc exec bash [20:36] then look in /var/log/cloud-init something in there [20:37] aemara: not at this stage, but it is on our roadmap [20:37] thumper: yes, one instance with state running and type persistent. [20:37] aemara: does GCE allow you to specify this after the instance has started? [20:37] but it has no ipv4 or ipv6 address [20:37] charlls: so it looks like the networking is not working [20:37] @thumper only if i shutdown the instance. I cant do it while it is running :(. [20:38] the cloud init output from the inside of that machine should tell us why [20:38] during bootstrap I get a line that says "Resolved LXD host address on bridge lxdbr0: 10.122.27.1:8443" [20:40] charlls: are you on a vpn or something that would try to proxy that address? [20:40] charlls: really info from inside the container is necessary to debug [20:40] lxc network show lxdbr0 shows ipv4.address: 10.122.27.1/24 [20:40] no, just behind my home router [20:40] charlls: does lxc list show an address for the container? [20:41] no [20:41] it seems to get stuck precisely getting an address [20:41] not sure how to debug the networking issue [20:42] charlls: can you get the cloud-init logs and pastebin them? [20:43] hmm, I don't seem to have cloud-init installed [20:43] charlls: 'lxc exec juju-xxx-0 bash' where juju-xxx-0 is replaced with the name in lxc list [20:43] that should give you a root shell in the container [20:44] awesome, let me try [20:44] concerning `network_get()` [20:44] charlls: then there should be two file: /var/log/cloud-init.log and /var/log/cloud-init-output.log [20:44] would this be the appropriate way to use network_get() https://gist.github.com/jamesbeedy/b3ee89d9c266fb374db41e4642505a0e#file-provides-py-L13 [20:44] charlls: ideally grab the content of those files [20:45] bdx: I'd have to defer to wallyworld or jam [20:45] thumper: I didn't mean to target you - sorry for interrupting :) [20:45] but yeah [20:45] bdx: that's fine, I just didn't want to leave you hanging [20:47] wallyworld, jam, wpk, cory_fu: should I be calling network_get() in my interface code or layer code? [20:47] I feel like I should be calling network_get() at the layer level, and passing the desired value returned from network_get() to the relation [20:47] charlls: also, you can cat the file from outside the container : lxc exec juju-xxx-0 cat /var/log/cloud-init-output.log [20:48] instead of what I have going on ^, where I'm passing the relation the interface_name and calling network_get() from the interface code [20:48] bdx: Yeah, that's what I would recommend as well. [20:48] yeah, I was wondering how to copy the content outside the container. Thanks [20:49] bdx: The interface layer should really only be concerned with providing a consistent API for the communication channel and dealing with any serialization or other transport concerns [20:49] cory_fu: thx [20:50] cory_fu: what are the implications of the new endpoint for the peer relation type? [20:51] bdx: It should work the same, and actually be more consistent. The only difference is that you can expect that there will only ever be one relation, so you can just use the all_units helper if you want [20:52] cory_fu: thats awesome!!!!! [20:55] charlls: also from lxd expert, the lxd version, contents of /var/log/syslog, and ps fauxww from the host [20:55] thumper: https://pastebin.com/yvKGrQnN [20:55] charlls: he mentioned that the most common problem is running a DNS server on the host [20:56] charlls: that was the cloud-init.log right? can we get cloud-init-output.log too? [20:58] just contains a single line: Cloud-init v. 0.7.9 running 'init-local' at Wed, 25 Oct 2017 20:58:04 +0000. Up 15 seconds. [20:59] yes the one on pastebin was cloud-init.log [21:00] oh... [21:00] that's weird [21:00] charlls: what about tye syslog and ps? [21:02] the /var/log folder in the juju container has the following contents: apt btmp cloud-init-output.log cloud-init.log dist-upgrade dpkg.log fsck lastlog lxd unattended-upgrades wtmp [21:02] no syslog [21:07] this is ps fauxww inside juju container: https://pastebin.com/YnAKdkXG [21:22] charlls: sorry, on calls just now [22:04] ok no prob [23:58] if someone feels in a helpful mood, I posted my issue here: https://askubuntu.com/q/969275/17002