[00:55] I have different physical servers i want to use for specific packages. Is there a way to link the MAAS name and setup to a specific machine in Juju? This would allow me to allocate the charms to the appropriate servers. [01:32] bloodice: We use MAAS tags for that === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [02:03] lazyPower: damn... elasticsearch is failing the install hook now [02:03] apt.cache.FetchFailedException: W:Failed to fetch http://packages.elasticsearch.org/elasticsearch/1.2/debian/dists/stable/main/binary-amd64/Packages 403 Forbidden [02:04] lazyPower: I hope it is intermittent and temporary [02:21] thumper: that looks like an error server-side, probably moved files or changed to https [02:21] not moved files, but changed paths === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === Guest8948 is now known as jam [07:47] https://askubuntu.com/questions/576500/juju-bundle-deploy-in-one-machine-in-lxc/579585#579585 [08:39] lazyPower: o/ [08:39] o/ seal [08:40] is there a particular order in which to execute the flannel-docker? [08:40] I was unable to communictae with other host [08:40] will post the order which I ran it in a sec [08:41] seal: the order dependencies were resolved in the latest iteration of the charm i do believe. [08:41] http://paste.ubuntu.com/9952491/ [08:41] * lazyPower looks [08:42] hmm, that looks correct to me - can you get me the output of `juju run --service docker 'ifconfig -a'` ? [08:43] sorry, docker-api, in your given example. [08:43] hmm, that looks correct to me - can you get me the output of `juju run --service api-docker 'ifconfig -a'` ? [08:43] sorry this the correct code http://paste.ubuntu.com/9952527/ [08:44] so, you've got a single docker host, and you've made the connections to ETCD and flannel - which should have setup the docker0 bridge. This is why i want to take a peek at ifconfig -a to see whats going on with the docker0 bridge adapter. [08:45] ok one sec will set it up quickly and ping you if you are still around [08:45] destroyed the last environment :) [08:59] ok, np seal === lazyPower is now known as lp|Metrics === liam_ is now known as Guest51703 [14:08] sinzui: src/bitbucket.org/kardianos/osext/LICENSE is wrong. It's been fixed upstream though. [14:08] :/ [14:08] sinzui: can you arrange to pull the latest in please, assuming that doesn't break any dependants? [14:09] I will [14:09] sinzui: it might be worth instituting a licence review policy before permitting changes to dependencies.tsv. [14:09] Thanks [14:09] rbasak, I think it will be after the fact, engineers change deps all the time. I think we need a report of what change for review every week [14:12] sinzui: OK [14:13] sinzui: this isn't a blocker. Logically I think it's acceptable as it's fixed upstrema, and I've written a note accordingly. [14:21] rbasak, I reported bug 1416425 and am handing it off to developers [14:21] Bug #1416425: src/bitbucket.org/kardianos/osext/LICENSE is wrong

[14:25] sinzui: thanks. Next... [14:25] sinzui: why has juju/cmd changed license? [14:26] sinzui: I don't mind particularly, but I do have to update the copyright file to follow all these changes, and it's not optimal for me to discover these all in the diffs. [14:26] It's driving me up the wall a little right now. [14:27] rbasak, I don't know about the change, sorry. I will ask [14:28] sinzui: it's not so much that it changed, but more that it feels like these changes are almost slipping past me. We should sort out a process so that I can follow licensing issues and changes more easily. [14:30] rbasak, Noted. I will bring this up as an issue in the release meeting in 2 hours. We will need a policy to catch and review licensing changes before we *consider* a release [14:31] Thanks. [14:39] sinzui: github.com/juju/syslog has an incorrect LICENSE file. [14:40] Google's third clause has been removed. [14:40] :( [14:41] Not fixed upstream. Can I leave this one with you? I think if it's fixed upstream it's OK to take it. [14:41] (without updating it, for now) [14:44] rbasak, I am confused https://github.com/juju/syslog/blob/master/LICENSE shows me a bsd 2-clause license. How does this relate to google? [14:47] sinzui: right, but for example https://github.com/juju/syslog/blob/master/syslog.go refers to the Go LICENSE which hasn't been shipped. [14:48] ah [14:48] thank you rbasak [14:48] sinzui: it seems that the author derived from Go, but just included some arbitrary LICENSE file instead of the Go LICENSE file. [14:48] No problem. [14:48] I'll keep slogging through. [14:49] sinzui: next question. https://github.com/juju/testing/issues/32 hasn't been addressed. Is this reported to the right place? [14:50] rbasak, possibly not. I report bugs against https://bugs.launchpad.net/juju-core when I want core engineers to fix the issue. I don't think the team reads github issues unless pressed [14:53] we are sedding some juju status output to get the instance id, and I was wondering if there is a nicer way to do that? juju status doesn't exist but would it be nice? [14:53] OK, I'll copy it over. [14:54] Copied to https://bugs.launchpad.net/juju-core/+bug/1416430, thanks. [14:54] Bug #1416430: Some files refer to an include license file that is not included

[14:55] sinzui: next issue. juju/utils introduced a bunch of files claiming AGPLv3, refering to a LICENSE file, when all other files are LGPLv3 and so is the LICENSE file. [14:55] Should these be LGPLv3? [14:56] I don't know, but I agree that LGPLv3 is right. [14:57] rbasak, I suspect someone moved code and didn't check the license when they moved it [14:57] Yeah that sounds likely. [14:58] I'm going to stop now. I'll carry on later. Probably next week. [14:58] * skay discovers --json and is much happier [14:58] * skay means --format not --json [15:01] rbasak, I will report a bug about the utils issue. For each licensing bug, I have target 1.23, 1.22, and 1.21, and we can see people are working on them https://bugs.launchpad.net/juju-core/+milestone/1.21.2 [15:09] rick_h_, hey - is there a way to use juju-gui in a do what I tell you mode [15:10] updating bundles that use lxc containers etc... is a pita right now as I appear to have to use MAAS environment todo the updates [15:11] sinzui: thanks [15:14] jamespage: not following. You mean without the commit stage? [15:14] rick_h_, I really just want a demo environment where I can tweak things [15:14] rick_h_, but the demo online appears to ignore machine placements [15:14] jamespage: you mean have it work like the demo url? [15:15] rick_h_, yup [15:15] jamespage: oh, hmmm, how are you defining machines? it should support what it exports. [15:15] jamespage: so to put it in the same mode there's a sandbox setting on the gui to set [15:15] jamespage: looking for exact config attr [15:16] jamespage: juju set juju-gui "sandbox=true" [15:16] rick_h_, this bundle - https://jujucharms.com/u/openstack-charmers/openstack/ [15:22] jamespage: so few things. One, that sandbox setting should get you into a 'demo mode' and the bundle.yaml you dump out from the gui should reimport with drag/drop. [15:22] jamespage: looking your bundle doesn't show in the old api so not sure what's up. looking into it [15:24] Is there a safe way to compact juju's monodb database? [15:26] jamespage: loading up for flight but if you need anything hit up hatch or frankban and they'll help out. === roadmr is now known as roadmr_afk [15:57] hatch, frankban: when I import/export that bundle through a gui in sandbox mode, I lose the machine placement - any ideas? [15:58] jamespage: sorry that's not yet implemented [15:58] hatch, its not implemented in the sandbox mode? [15:58] jamespage: the GUI does not support machine placement in bundles yet [15:58] frankban, really? [15:59] frankban: I thought that was only in sandbox mode [15:59] frankban, but I created this bundle on a MAAS deployment using juju-gui ? [16:00] hatch: oh, so we have machine placement (as defined by the new bundle spec) in real env in the GUI? [16:01] hmm - I'm not actually confident on that now :) [16:01] let me look at the source [16:02] hatch: thank you [16:04] frankban: looks like it just passes the bundle file through so as long as the deployer supports it... [16:04] jamespage: is it not working for you in a real env or just in sandbox mode? [16:04] hatch, it worked ok on a MAAS environment (import/export) [16:05] hatch, but that's a pita for doing a small change to a bundle [16:05] that's a good point - you are able to modify the file manually though [16:05] hatch, but not for sandbox - as soon as I import the bundle it looses its placement data [16:05] hatch, I was able to modify the file manually [16:05] yeah sandbox is definitely not supported for those machines [16:05] er machine placement [16:06] hatch, gui doesn't export placement data? [16:07] hazmat: it doesn't support placement data on import in sandbox mode. I'm just tracing the export code now (it's been a bit since I've been in there) === roadmr_afk is now known as roadmr [16:11] hazmat: if I'm reading the code correctly it should be exporting machine placement data as well [16:12] jamespage: so you would like to import a bundle file into sandbox so that you can modify it without hacking on the yaml file directly? [16:22] hatch, +1 yes please [16:23] jamespage: would you be able to create a bug to that effect? [16:26] hatch, I can do that [16:31] jamespage: thanks! [16:32] hatch, its annotating x/y gui placement, but machine placement? [16:33] I don't understand the question [16:34] hatch, its not exporting unit to machine placement, just service x/y on canvas [16:36] hazmat: ok I'll have to look into that because according to the export code it should be exporting the machine placement details [16:37] hatch, not sure how, unless its storing placement as annotation, that info is loss and would have to be inferred. anyways, not seeing that on demo.jujucahrms.com [16:37] hazmat: jamespage we decided not to implement deployer logic in JS for importing with placement hoping we were going to get core support and skip that. === kadams54 is now known as kadams54-away [16:37] hazmat: https://github.com/juju/juju-gui/blob/develop/app/models/models.js#L2320 [16:38] hazmat: jamespage but as priorities go that's a decision from last sprint we should look at again as it's something we should give users [16:38] export is ok [16:39] jamespage: correct, and atm there's a bug in export hatch is looking at. I think we missed an update as we've upgraded to the new charmstore api and that should be a quick fix hopefully. [16:39] hatch, interesting, might be a bug, that looks like its trying to infer it, but its not triggering on demo site in sandbox [16:42] hazmat: ok I'll be sure to check that out - I'm now fixing an export bug which was just discovered === kadams54-away is now known as kadams54 === roadmr is now known as roadmr_afk [18:18] I have different physical servers i want to use for specific packages. Is there a way to link the MAAS name and setup to a specific machine in Juju? This would allow me to allocate the charms to the appropriate servers. === roadmr_afk is now known as roadmr === lp|Metrics is now known as lazyPower === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [21:38] I have different physical servers i want to use for specific packages. Is there a way to link the MAAS name and setup to a specific machine in Juju? This would allow me to allocate the charms to the appropriate servers. [21:38] bloodice: --to option [21:39] That lets me select the machine, but there doesnt seem to be a way for me tell juju when i create a machine which server to take from MAAS [21:40] bloodice: when you juju bootstrap? [21:43] well, when i bootstrapped, it just grabbed a server [21:43] bloodice: yep do juju bootstrap --to [21:44] but after the gui is up and running, there doesnt seem to be a way to say "i want this specific server in MAAS" [21:44] not sure about the gui, but cli you use --to [21:44] so after Juju is online, i can go into the console and create a machine with the --to command using the maas hostname? [21:45] bloodice: yep === kadams54 is now known as kadams54-away [21:46] hrm === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away === zz_CyberJacob is now known as CyberJacob [22:43] i ran this command "juju deploy --to c7yra.maas mysql" and i get: invalid --to parameter "c7yra.maas" [22:44] i tried "juju add-machine --to c7yra.maas" and got: error: flag provided but not defined: --to [22:53] ack, ok, i got it to work, but i really dont like it... i needed to use the tag feature, so "juju add-machine --constraints tags=c7yra.mass" [22:55] it just adds a machine with a number, but i have no way of seeing that 8 is server c7yra.mass in the gui. Status does show the server name though. [22:57] welp, that doesnt work anyhow... status says: ({"tags": ["No such tag(s): ''c7yra.maas''."]})' [23:08] can any one help with JuJu installation I am getting not able to connect: Permission denied (publickey,password). [23:19] just passing to drop something I did the other night for Juju + Vagrant https://github.com/sebas5384/vju