[04:11] Has anyone been able to successfully deploy the neutron-gateway using 16.04? I've tried using a global kernel parameter of "net.ifnames=0" in the MAAS global config, but i'm still failing at the same step during the autopilot deployment. === frankban|afk is now known as frankban [07:54] bildz, which juju version and which maas version [07:55] rc1 has some challenges with automatic bridging of unconfigured network interfaces [07:59] marcoceppi, hey - can I request a point release of charmhelpers? I added a get_upstream_version helper to host a while back, which we're using in legacy charms that still sync [07:59] but I need to have it for new reactive charms as well pls :-) === rogpeppe1 is now known as rogpeppe [09:20] what does juju-info relation provide? I see you can use it for subordinate charms and get private addresses, but is that it? [12:49] spaok: Its reason for existence is to support subordinate charms. You could probably stuff extra information on there, but you really should declare a real relation instead of abusing juju-info for that. [14:50] jamespage: you got it [14:52] marcoceppi, ta [15:01] i'm fooling about with juju and lxd again today.. following a reboot, i cannot get more than 10 containers to start up [15:01] jamespage: good morning [15:01] saw your response last night/morning [15:02] the 11th container start hangs with forkstart running [15:02] ii conjure-up 2.0.1~beta2-0~201609281246~ubuntu16.04.1 all Package runtime for conjure-up spells [15:02] ii juju-2.0 2.0-rc2-0ubuntu1~16.04.1~juju1 amd64 Juju is devops distilled - client === frankban is now known as frankban|afk === rogpeppe is now known as rogpeppe1 === rogpeppe1 is now known as rogpeppe === rogpeppe is now known as rogpeppe1 === rogpeppe1 is now known as rogpeppe [17:18] Quick question: Juju 1.25 appears to be able to target vSphere clouds, did that support disappear with 2.0 or not documented yet? [17:19] nicbet: it still exists with 2.0 [17:20] nicbet: just seems to be suspiciously missing documentation [17:20] larrymi_afk: could you share what you did to configure vSphere with juju 2.0? nicbet is asking [17:20] which the reactor charms, how do I trigger on a hook? I've tried a few different patterns I've seen in charms, but when running the charm it seems to always skip the hook [17:21] marcoceppi: thanks! looking forward to larrymi_afk's pointers. [17:26] marcoceppi: nicbet: sure [17:27] nicbet: you will need an entry in your clouds definition pointing to your vcenter. [17:27] nicbet: like this [17:27] vsphere: [17:27] type: vsphere [17:27] auth-types: [userpass] [17:27] endpoint: [17:27] regions: [17:27] dc0: {} [17:28] nicbet: for endpoint, you'll specify the IP of your vCenter [17:29] nicbet: in the credentials, you'll use auth-type: userpass [17:29] then spedify password: and user: that you used to access the vCenter [17:30] larrymi_afk: excellent, so it appears most of the 1.25 documentation transfers to 2.0, with the change to auth-type: userpass [17:30] nicbet: that's pretty much it .. when you bootstrap your controller, juju will ask you export an environment variable.. then once you export it, then everything is smooth sailing [17:32] nicbet: yes, not much is different.. it's been months since I used 1.25 so I don't recall exactly what I had in my environments.yaml and I think I passed a few arguments through the command line like the datacenter [17:33] other than that, things are relatively similar [17:35] nicbet: I didn't have a regions definition either (just the datacenter that I passed on the command line).. but 1.25 may have changed to be more in line with 2.0 since.. [17:42] larrymi_afk: thanks, with the details you provided, I created the yaml file, bootstrapping the environment now... [17:43] nicbet: cool === larrymi_afk is now known as larrymi_ [18:06] larrymi_ : no go. when bootstrapping the controller the provider brings the controller node online in vCenter, obtains IP address and then fails on SSH. [18:07] nicbet: what's the error? [18:08] larrymi_: ERROR failed to bootstrap environment: waited for 10m0s without being able to connect: Permission denied (publickey). [18:08] nicbet: looks like maybe cloud-init failed === fginther` is now known as fginther [18:09] nicbet: trusty or xenial? [18:10] larrymi_: xenial [18:11] Hello Team, Not able to login using bzr launchpad-login "user"... getting error bzr: ERROR: Connection error: Couldn't resolve host 'launchpad.net' [Errno -3] Temporary failure in name resolution [18:11] can you please advise me on this? [18:12] larrymi_ : juju help bootstrap says: "Private clouds may need to specify their own custom image metadata, and [18:12] possibly upload Juju tools to cloud storage if no outgoing Internet access is [18:12] available. In this case, use the --metadata-source parameter to point [18:12] bootstrap to a local directory from which to upload tools and/or image [18:12] metadata." [18:16] nicbet: there was a cloud init issue in xenial that was fixed recently.. not sure if you're pulling an older image. Try with trusty to see whether you run into this. btw, there's a DNS issue with Xenial that you could run into after deploying but shouldn't impact the bootstrap. [18:16] older image than the fix [18:17] larrymi_ how do I specify which OVF image juju deploys? [18:22] nicbet: you could try to bootstrap by setting the image-stream.. for example, --config image-stream=daily [18:23] not sure about what other values are available for image-stream or if you can point to a url [18:26] how do I get the remote unit's ip info from a relation? when I try hookenv.relation_get('private-address', unit, relid) I get None [18:36] generic question... looking at https://jujucharms.com/docs/1.25/storage, what is the definition of a pool and a volume? how do they relate? [19:10] larrymi_ : https://bugs.launchpad.net/juju/+bug/1588041 is what I'm running into. [19:10] Bug #1588041: [2.0 rc1] juju can't access vSphere VM deployed with Xenial, cloud-init fails to set SSH keys

[19:25] nicbet: ah it should be fixed though. Have you tried daily? [19:25] larrymi_: yes didn't work for me I'm now trying to figure out to get debug logs so I can inspect which image is being pulled [19:27] nicbet: if you use --debug, I think that should have the info for where it's getting it from [19:28] howdy... i seem to be having trouble using a 4k ssh key with juju... 2.0-rc2-elcapitan-amd64 [19:30] is was reported fixed in https://bugs.launchpad.net/juju/+bug/1543283 but not seeing the same [19:30] Bug #1543283: [Joyent] 4k ssh key can not be used: "cannot create credentials: An error occurred while parsing the key: asn1: structure error: length too large" [19:31] larrymi_ : ova_import_manager.go:175 Downloading ova file from url: http://cloud-images.ubuntu.com/daily/server/xenial/20160930/xenial-server-cloudimg-amd64.ova which is the image with cloud-img fixed [19:35] nicbet: strange, I have a controller from yesterday that worked fine.. I'll redeploy to see. [19:36] nicbet: which version of vsphere? [19:39] larrymi_ 6.0-20160104001 [19:47] the bug is marked as "Fixed released", but i'm not seeing the fixed behaviour [19:54] nicbet: worked for me for http://cloud-images.ubuntu.com/releases/server/releases/xenial/release-20160922/ubuntu-16.04-server-cloudimg-amd64.ova which is build where the fix would have gone in... now trying daily [19:55] nicbet: which juju version are you using? [20:01] nicbet: worked with the 9/30 build too. [20:21] can can open-ports be gotten via hookend.relation_get? I know privateaddress works, but I get None for open-ports [20:22] trying to make a subordinate charm that gets the ip and ports for a service [20:57] I have a general question about isntances of charms. If the software the charm is written around requires physical access to a device how do you deploy multiple instances and configure each instance to use the specific physical device per machine? For instance, ceph-osd. Each OSD will connect to a different “physical” drive. If I have 4 servers and the “physical” drives are not the same on all 4, how do I configure ceph-os [20:57] use the proper device on the corresponding machine? [21:06] valeech: so in this case there's a "leader" idea in an application to help coordinate among units and so using the leader hooks would help keep track of who's on which devices/etc. [21:15] rick_h: makes sense! I just read up on leadership hooks. ty! [21:37] larrymi_ : version is juju 2.0-rc2-xenial-amd64 [21:41] nicbet: it matches what I have [21:42] nicbet: try trusty to see if it's able to bootstrap completely [21:43] juju bootstrap --config default-series=trusty [21:46] larrymi_ will try that next. every attempt takes quite a bit of time to push the ova images - it doesn't seem to cache the downloads on fail [21:47] nicbet: ok.. if you look at comment 18 then this is the bug where the fix went in. [22:07] larrymi_ who's prividing the cloud-init metadata to the box? juju? [22:14] nicbet: idk [22:22] am i correct in guessing that reactive set_local requires that the value be json serializable? I was caught by surprise when python error'd out [22:42] larrymi_ : according to https://github.com/juju/juju/blob/master/provider/vsphere/ova_import_manager.go line 72 the cloud init params are given as part of the OvfImportSpec