=== freeflying_away is now known as freeflying === defunctzombie is now known as defunctzombie_zz === defunctzombie_zz is now known as defunctzombie === defunctzombie is now known as defunctzombie_zz [01:22] Is there a way to modify the config of a charm from within a hook? [01:23] ZonkedZebra: no, there is no config-set hook command [01:24] davecheney: best approach to auto load new code from a remote repo? cron? hooks? config bool that you set back and forth? [01:25] ZonkedZebra: why not [01:25] juju set revision=XXX [01:26] which will fire the config-changed hook on your units [01:26] davecheney: if I juju set to the value it already is does that trigger the config-changed hook? [01:27] (The goal is to have production and staging that track the appropriate git branch with minimal intervention) [01:29] ZonkedZebra: no [01:29] Buuuuuut, remember as a charm author, we do not guarentee that hooks will not be run only once [01:29] so they should be written to expect this [01:31] ZonkedZebra: yep, thats fine. I've got all the appropriate checks to make sure multiple runs don't cause issues. Just looking for the easiest way to poke it a little to have pull the new changes from git [01:32] ZonkedZebra: why do you want config-set (sic?) [01:32] it smells like you are trying to tell someone else that the charm changed something [01:35] i just created a config property that I would going to set to true, and then reset to false in config-changed [01:36] what would it resetting to false mean ? [01:36] That i could consistently do "juju set app pull=true" [01:36] i'm not trying to troll btw, just trying to understand your problem to fit it into the (sometimes limiting) model that Juju offers [01:36] what does pull=true do ? [01:37] ie, why not just setup a cron on the unit ? [01:37] it would fire config-changed (where git pull happens) and then set back to false to be ready to get pull=true set again [01:38] ZonkedZebra: so, something would be cron'd on the client to setup pull=true ? [01:38] why not just setup the cron on the unit ? [01:38] No, I would do that by hand [01:38] ok, in that case i'd recommend [01:38] not pull=true [01:38] do work, commit, do work, commit, juju set app pull=true [01:38] but revision=XXXX [01:39] Would also work, but I would like the units associated with branches, not a particular rev, and as we discussed, if branch is already dev then set branch=dev would not trigger the pull [01:40] sure, have two config setings [01:40] branch=... rev=... [01:40] if you just need a trigger [01:41] juju set app rev=$(pwgen 100) [01:41] just to set it to some random garbage [01:41] * ZonkedZebra nods [01:41] davecheney: that will do, thanks [01:41] np [01:42] pwgen 100 may be overkill [01:42] maybe call the config valye nonce or trigger or something [01:45] davecheney: timestamp so at least it will be slightly useful [01:47] suredate +%N [01:47] date +%N [01:47] maybe [02:03] davecheney: hi, I use juju add-machine and OS installation failed due to some post install script failure. now I can't juju terminate-machine. the life-cycle is dying... what can I do here? [03:07] hey marcoceppi, is it possible to get the postfix charm on the store before oct 1st? [03:17] julianwa_: i'm sorry you got bit by this [03:17] this is an open bug/freature request [03:18] your best bet is to delete the machine using aws or whateve ryou use [03:18] then ignore the broken record in the juju status [03:18] it's been a known issue for a long time [03:18] i'm trying to get it bumped up th epriority list [03:18] but please don't take that as a forward looking statement [03:21] davecheney: you mean leave the dying machine there? but the dying server will have same maas-name. is that ok? [03:24] oh, you're using maas [03:24] hmm [03:25] lol [03:25] julianwa_: how did the machine get killed ? [03:25] did you use mass to kill it when the terminate-machine failed ? [03:26] davecheney: not killed. one post-install script failed when juju add-machine. Then I execute terminate-machine [03:26] davecheney: it's still in MAAS [03:27] julianwa_: are you Canonical ? [03:27] davecheney: yes... [03:27] lets talk in that other channel [03:27] sorry folks, i'll post a wrap up [03:27] when I figure out the problem === freeflying is now known as freeflying_away === freeflying_away is now known as freeflying === defunctzombie_zz is now known as defunctzombie === freeflying is now known as freeflying_away === freeflying_away is now known as freeflying === defunctzombie is now known as defunctzombie_zz === freeflying is now known as freeflying_away === freeflying_away is now known as freeflying === freeflying is now known as freeflying_away === CyberJacob|Away is now known as CyberJacob [07:30] marcoceppi, jcastro, ping === freeflying_away is now known as freeflying [08:31] hi, I've added a new unit to an existing juju deployment and when I do a juju status all the other machines report in fine but the new unit reports "agent-state-info: '(error: failed to list contents of container: juju-stagingstack-geonames" [08:31] I can query the juju-stagingstack-geonames" [08:32] bucket fine and have downloaded bootstrap-verify, provider-state and the tools tgz without a problem [08:32] I tried removing the unit and adding another one but I get the same error === CyberJacob is now known as CyberJacob|Away === freeflying is now known as freeflying_away [10:30] heya all - I've been getting "cannot log in to admin database" immediately after a bootstrap on Openstack [10:30] this is with 1.14.1 on raring [11:54] fwereade: pong [11:55] marcoceppi, hey, I was pinging you as a possible evilnickveitch proxy, but I see he's online now [11:55] gnuoy: I've not seen that error before, what provider are you using? What version of Juju? [11:55] fwereade: ack [11:56] evilnickveitch, ping [11:56] fwereade, hi [11:56] marcoceppi, openstack and 1.13.2-1 bzr revno 1670 [11:56] bloodearnest: is this /immediately/ after bootstrap, or after you can verify that the bootstrap is running via juju status [11:56] evilnickveitch, I was wondering if there was anything I could do to ease the passage of the docs I gave you a whileback? [11:56] marcoceppi, what object is it trying to access ? [11:56] evilnickveitch, a casual look seemed to indicate they weren't up yet [11:56] gnuoy: I have no idea, did the machine come online? [11:57] evilnickveitch, if the problem is, say, that they're crap, I'd like to help make them less so :) [11:57] fwereade, oh, thanks for the offer - they are not, because I have taken the opportunity to include them in more of a restructure, but they will be up later today [11:57] marcoceppi, nova thinks its active, I'll try ssh'ing to it [11:57] evilnickveitch, ok, that's awesome, tyvm [11:57] gnuoy: if you can ssh in to it, then you can get the /var/log/juju/machine-*.log file - should help shed some light [11:58] fwereade, no, they aren't crap at all :) [11:58] I will let you know when they go up, would be good to get your feedback [11:58] evilnickveitch, jolly good, i felt obliged to check ;p [11:58] evilnickveitch, cheers [11:59] marcoceppi, well it looks like I was looking in the wrong machine and the new machine never came. I'll have a dig around and see if I can see why. thanks [12:01] gnuoy: also, you might want to consider moving to 1.14.1 as it's the latest "stable" [12:01] marcoceppi, absolutely [12:01] I noticed the environment was named "staging", but the latest dev will be 1.15 so 1.14.1 is the truely latests [12:02] it should also be easier to upgrade between stable versions than dev releases when doing in-place upgrades [12:02] * marcoceppi is so giddy about in-place juju upgrades [12:04] marcoceppi: I get that running juju status [12:04] marcoceppi: it times out after about 7min with "Unable to connect to environment "openstackshredder"" [12:05] bloodearnest: that's interesting. Can you destroy then bootstrap again with `--debug -v` options, then run `juju status -v --debug` [12:05] marcoceppi: ack [12:05] bloodearnest: also, you have admin-secret set, correct? [12:05] bloodearnest: in your environments.yaml [12:06] marcoceppi: yep - freshly generated with generate-config [12:06] bloodearnest: excellent, if you could pastebin those when you get them that should help shed some light [12:09] marcoceppi: destroy fails: https://pastebin.canonical.com/98138/ [12:09] marcoceppi: the --debug points at opendns issues [12:10] some kinda redirect issues [12:10] bloodearnest: that's annoying [12:11] marcoceppi: yeah - am trying from a canonistack instance I use for dev, but I'm having similar problems there too [12:24] marcoceppi: bootstrap output: https://pastebin.canonical.com/98139/ [12:25] marcoceppi: status output: https://pastebin.canonical.com/98140/ [12:26] bloodearnest: yeah, it's successfully connecting to the bootstrap, just not logging in for some reason :\ === bing is now known as Guest14737 [12:27] marcoceppi: for completeness, destroy output (3 lines): https://pastebin.canonical.com/98141/ [12:27] all this is done from a nother vm on the same OS environment [12:28] bloodearnest: I've not encountered this, not quite sure how to debug past here. You might find more information on the bootstrap node in /var/log/juju/ [12:35] marcoceppi: don't know if it's related, but I can't ssh into the boostrap node - publickey denied [12:35] bloodearnest: well, that's also interesting [12:41] marcoceppi: in another env, I seem to be able to ssh in === defunctzombie_zz is now known as defunctzombie === freeflying_away is now known as freeflying [14:15] Assuming i joined the right ubuntu catch up..... hello geeks nerds and all :) === defunctzombie is now known as defunctzombie_zz [16:40] marcoceppi, charm-tools uploaded to saucy - got accepted [16:40] hi jcastro i just send mail about bluemind and juju ;-) [16:40] jamespage: \o/ Thank you! [16:41] marcoceppi, hey np === cmagina is now known as cmagina-away [16:53] jamespage: the next step is to get charm-tools in to backports for precise, when I run requestbackport it says no published binaries in saucy. Is this just a waiting game? [17:01] Charm Championship submission charm school on http://ubuntuonair.com in a few minutes! [17:08] marcoceppi, give it a chance to get into the release pocket [17:09] jamespage: ack, figured [17:09] jcastro: you need me there, or is this a Mims and you thing? [17:11] Hello were are getting kicked off on the charm school, "How to enter the Charm Championship." [17:11] jcastro: oh bugger. The new package removes charm-helper-sh, which is provided in saucy. I suppose that's going to be a problem during the backport req [17:11] jamespage: ^^, not jcastro [17:12] bah, that whole sentance is wrong [17:12] jamespage: oh bugger. The new package removes charm-helper-sh, which is provided in the precise version. I suppose that's going to be a problem during the backport req [17:14] if you would like to follow along for the charm school it is at http://ubuntuonair.com/ [17:15] YouTube direct link is @ https://www.youtube.com/watch?v=c6wTtWDyXsc [17:24] sound went out completely :-( [17:24] I can't hear anything... gonna try to reconnect... sorry for the technical difficulties [17:40] hi === cmagina-away is now known as cmagina [19:36] has anyone else been trying to setup HA openstack with Juju? I've been following the guides posted and have my setup 90% there.... instances are launching and running but are not getting an IP from quantum at all. When I check the logs they just show APMQ messages succesfully crossing. anyone have similar issues? === CyberJacob|Away is now known as CyberJacob === defunctzombie_zz is now known as defunctzombie [20:33] zradmin: a few people have been setting up openstack, let me see if I can recall their names [20:38] marcoceppi: thanks! [20:38] kurt_: were you the one working on deploying openstack? [20:39] yup === defunctzombie is now known as defunctzombie_zz [20:43] kurt_: did you ever get far enough to experience quantum not assigning IP addresses? [20:43] I'm finished and it all works for me. [20:43] I had to manually configure quantam. [20:45] marcoceppi: I could never get the charm to work out of the box [20:45] I only allowed the charm to do the basic install, but did all post-configuration myself [20:54] kurt_: what was the post configuration? the charm i specified eth1 as extnet but it looks like something in juju changed so it uses alot of lxc bridges [20:54] kurt_: I have nodes with 2 nics, one on the "internal" switch and one on the "external" switch [20:55] right - one nic should connect to your oam net, the other to you external lan [20:55] zradmin - do you do evernote? I've put it all in to that format so you can see [20:56] kurt: not currently but I can create an account real fast [20:56] do that and I'll share the note [20:56] kurt: evernote username is zradmin :) [20:58] k, hang on a sec [20:59] Actually you may not need account [20:59] see if you can see this [20:59] https://www.evernote.com/shard/s244/sh/37674b81-51af-4579-9579-8058b4cf3a9a/aca1835adea4ca6cb52e7d0091ced91c [20:59] yup got it [21:00] There you go [21:00] that should answer your questions [21:00] thanks, I'll let you know how it turns out :) [21:00] good stuff [21:01] FYI - the process was shamelessly borrowed from Kentb on the security team === defunctzombie_zz is now known as defunctzombie [21:30] kurt_: hmmm it looks like it does the same thing I was doing in horizon to configure the ext_net, I created a new project and created the networks via commandline and am still having the same issue [21:32] kurt: this is what I get on the instance http://pastebin.ubuntu.com/6164559/ [21:44] zradmin: is your ext_net hooked up to a separate network? [21:44] it looks like its trying to bring it up on eth0 instead of eth1 too [21:45] yeah it is set to configure to eth1, but when I do an ifconfig on the nova-compute node it doesnt show anything configured on eth1 [21:45] you need to specify eth1 for the quantum charm [21:45] yeah i did that [21:46] no ip address [21:46] this is in the syslog on the node m7q49 dnsmasq-dhcp[2221]: DHCP packet received on qvo6a26ce05-ae which has no address [21:46] are you certain your eth1 is alive and connected to a network other than your eth0? [21:47] are you doing this with physical hosts or virtual hosts? === defunctzombie is now known as defunctzombie_zz [21:48] physical, building on an m1000e blade [21:49] so ensure your eth1 is actually wired to a second network. you may need to test that part - because I think that's where your problem is [21:49] also - are you specifying precise:grizzly? [21:50] here - have a look at my local.yaml - make sure yours is similar [21:50] http://pastebin.ubuntu.com/6164613/ [21:51] mine goes for a single node rather than multinode installation [21:51] well, let me rephrase... [21:51] I am going the non-HA route for now as proof of concept [21:51] I installed on 6 virtual hosts [21:52] ah i see it, my switch stack is messing up my vlans [21:52] its tagging the traffic on the port [21:53] ;) [21:54] I need to take off for a while - I'll be back in an hour or so. [21:54] ping me if you are still having problems after you figure out your tagging problem [22:02] <_mup_> Bug #1232282 was filed: maas provider: bucket download failures not handled well === defunctzombie_zz is now known as defunctzombie [22:16] Is there a one liner to obtain the public address of a unit? [22:21] ZonkedZebra: uh, kind of [22:21] juju status api/0 | grep public-address | cut -d ":" -f 2 | tr -d " " [22:21] Something like that? [22:21] basically, I would have used awk, but that's because I <3 awk [22:21] * ZonkedZebra is not an awk user [22:21] awk cleaner? [22:22] ZonkedZebra: that's highly subjective :P [22:22] what would it be in awk? [22:24] ZonkedZebra: juju status wordpress/0 | grep -m1 public-address | awk '{print $2}' [22:24] -m1 is to only do the first match, in case there are subordinates [22:24] Worth learning I guess, After all, everything has awk [22:25] ZonkedZebra: awk is it's own language, but at the surface it's pretty easy to use [22:25] it's awesome for one-liners. beyond that I lose interest, hehe [22:25] it's like vi/vim. You learn one basic flag and life is good, then when you need to dig deeper, you can [22:26] (that flag is -F) [22:26] lol [22:26] yes, -F is awesome. :) [22:28] I only learned of awks amazing depth about two years ago, up until then it was my go-to "cut" === CyberJacob is now known as CyberJacob|Away === defunctzombie is now known as defunctzombie_zz === freeflying is now known as freeflying_away [23:08] see + awk are awesome [23:08] sed + awk are awesome [23:09] old as the hills but still as good as gold [23:14] This is a new error for me, "error: no relation id specified". Seems to be triggered by a call to relation-get. Ideas? [23:15] If only the hook tools had real documentation.... [23:18] Probably because I'm calling a script shared by multiple hooks [23:19] Whats the common practice for sharing code/functionality between hooks? [23:33] marcoceppi: ping [23:38] kurt_: i fixed the switch issue and restarted the compute node... but it still dosn't seem like traffic is going accross the bridge here's my interfaces: http://pastebin.ubuntu.com/6164893/ [23:40] zradmin: you've got a whole lot more on my quantum-gateway than I do [23:40] did you use my local.yaml as your deployment template? [23:41] here are my interfaces [23:41] http://pastebin.ubuntu.com/6164899/ [23:42] yeah my settings for those charms match up [23:43] hmm I dont have a br-ext [23:53] again - you must be having some issues with your eth1 [23:53] a bridge can't be created [23:54] look for hints in /var/log/syslog or dmesg [23:54] yeah im still looking into it... i hope something isn't wrong with that test node [23:54] try to manually create the bridge [23:54] ty [23:55] once you can manually create the bridge, you should be golden [23:55] maybe you have some spanning tree issues? [23:58] zradmin: look in dmesg to see what's happening - look at my entries around br-ex [23:58] http://pastebin.ubuntu.com/6164928/