=== vladk|offline is now known as vladk === vladk is now known as vladk|offline === CyberJacob|Away is now known as CyberJacob === vladk|offline is now known as vladk [08:33] axw, thanks for squashing bug#1296739 [08:33] <_mup_> Bug #1296739: Cannot destroy lxc env on trusty with juju-core 1.17.4-0ubuntu2

[08:34] gnuoy: you're welcome === CyberJacob is now known as CyberJacob|Away === jaywink_ is now known as jaywink === vladk is now known as vladk|lunch === vladk|lunch is now known as vladk [13:32] small bug fix MP for gunicorn charm, if any ~charmers are free [13:32] https://code.launchpad.net/~bloodearnest/charms/precise/gunicorn/upgrade-path/+merge/213236 [13:38] I have a juju dilemma... I'm deploying a service with service config ala deploy --config /path/to/service/config to set some config options unique to this deployment [13:39] These seem completely lost when I go to do a juju upgrade-charm [13:39] Is there something I can do so upgrade-charm uses the same service config used to deploy the app to begin with? [13:39] hi bloodearnest, we are really swamped right now on the review queue, we will get to that one as soon as we can. [13:40] hi timrc can you do a juju get charmname to see the config values [13:40] mbruzek: sure, just an fyi. fwiw, without these changes, the current gunicorn charm can't cleanly upgrade from from older versions [13:41] bloodearnest, noted and we will try to get to it asap. [13:41] mbruzek, That returns the config opts with the values I've set, but these seem to be the default values (defined in the config.yaml that ships with the charm) in the context of an upgrade-charm.... I'll try again to confirm that though [13:41] timrc, Is this a charm that you are working on or one in the store? [13:42] mbruzek, When I debug-hooks'd the upgrade charm hook and did a config-get on the command line I was not getting the values I set, but the default values (e.g. all empty) [13:42] mbruzek, No, not at the moment [13:42] timrc, that is interesting that config-get is not returning the configured values. [13:43] mbruzek, I'm using Trusty/juju-core/local 1.17.5 [13:44] I was going to suggest that the charm is not handling the configuration options correctly but if you can confirm that config-get is not returning the value that you see from juju get charmname I would open a bug [13:44] mbruzek, Okay. Thanks [13:45] sorry timrc I have not experienced this problem first hand I can not think of what is causing the problem. [13:49] timrc: 1.17.7 is latest, if you can try using that release. If you get the same experience then that's a bug [13:49] marcoceppi, Thanks, I'll upgrade and try again [14:05] morning all [14:06] lazyPower: morning [14:06] o/ [14:08] got a question about manual provider [14:10] go for it [14:14] so I need a vm that juju is installed on? [14:15] then new vm’s that I provision jsut get a vanila ubuntu server installed on them? [14:22] Valduare, will juju deploy ubuntu work for you? [14:23] mbruzek: I have right now a freenas box for storage serving iscsi to an esxi host [14:23] Valduare: https://juju.ubuntu.com/docs/config-manual.html [14:23] and no idle hardware laying around suitable to setup something new yet [14:23] take a look, there's a few dependencies on the host being allocated with the manual provisioner [14:24] Valduare, I misunderstood your question. [14:24] s/provisioner/provider [14:25] ok [14:27] so I need to start out with a new vm to install juju on - im reading about it downloading some tools and such, how big of a virtual drive should I give it [14:28] 8gb should be plenty [14:30] ok [14:31] looking forward to getting some new hardware to get this all going propper heh [14:31] :) [14:31] Since you were asking yesterday i started setting up a MAAS lab on my server using virtual machines [14:32] i still think manual provider is the way to go if your not treating bare metal like commodity hardware [14:32] but my vote may change [14:34] it seems like it’d be nice [14:40] i'll let you know my findings [14:41] follow up with me on Monday :) [14:46] ok [14:47] gah my windows computer wont let me vnc into it - so cant access vsphere client [14:47] have ot drive over... [14:47] this is exactly why I want a new setup lol === vladk is now known as vladk|offline === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [16:22] marcoceppi, There is an interesting problem... installing juju-core and running juju init will create an environments.yaml which creates a local provider by default, but if you do not have juju-local installed, you run into problems. Maybe a local provider should not be provided by default by juju-core? [16:23] well, i think juju-local should be installed with juju-core, personally [16:24] Also if you upgrade juju-core and to forget to upgrade juju-local you again run into problems (or at least have the potential too). bootstrapping after I upgraded juju-core but not juju-local hung juju bootstrap and put me into a weird state that required I delete ~/.juju/local/ [16:24] marcoceppi, I wouldn't mind that [16:42] anything special I need to put for the juju hostname === roadmr is now known as roadmr_afk [16:58] can I deploy a precise machine from a different version where run the client? === hatch__ is now known as hatch [17:01] I'm a bit confused about how the version is of new machines is made [17:04] hmm am I doing this wrong, am I suppose to do sudo add-apt-repository ppa:juju/stable on a vm in my dc or on my workstation ? [17:05] Valduare on your workstation [17:05] oh! [17:05] I had just sat here firing up and provisioning a new vm for this hah [17:06] ok so how will it find the vm’s when I am at home and want to manage something [17:06] does it connect to the vm’s through ssh? and port forwarding etc. [17:06] im going to be doing manual provider [17:07] Valduare: check the doc link i posted earlier it covers that [17:07] you will be provisioning a new vm, and it has to have a user that supports passwordless sudo [17:07] then you register that vm with juju and it takes care of the rest. [17:08] overm1nd: i dont understand what you are asking [17:08] overm1nd yes - you need to specify the series when running `deploy` if it's different than the default configured series [17:08] lazyPower ^ I guessed :-) [17:08] so i’ll provision them and get them setup with nat forwarding of ssh port and connect juju to them through that [17:08] so I can connect at home or locally [17:09] hatch: cuz you're the man :) [17:09] lol [17:09] Valduare: sounds good 2 me [17:09] thanks hatch [17:10] np [17:18] hmm juju switch manual [17:18] ERROR "manual" is not a name of an existing defined environment [17:19] Valduare: edit your environments.yaml and change "null" to manual [17:19] at one time the manual provider was called null [17:19] ah I see “null:” in there [17:20] what is bootstrap-host [17:23] the explination does not tell me much :P [17:26] Valduare simply change the text [17:28] i mean - what is the bootstrap-host ? === roadmr_afk is now known as roadmr [17:38] anyone? [17:40] from what I can tell, bootstrap host is the host that gets installed first. The one with ID 0 in juju status [17:41] also known as the state machine [17:44] ah [17:44] so i can feed it the ip of the first vm I want to provision with it then [17:45] the host box is remote to me [17:45] is this environments.yml connecting throuhg ssh? [17:57] well this isnt working [17:57] gah [18:03] marcoceppi: hello [18:03] hello [18:28] does bootstrap-host: simply put an ip address with port of how to connect to the server via ssh? [18:28] what is it doing it does not have good explinations === Valduare_ is now known as Valduare [18:47] my machine links are all messed up after I disassociated my elastic ip on AWS from one server and associated with another. Anyone ever see this? [18:50] so do I change “null:” to manual: in the environments.yml [18:52] blah weblife comp pooped get you a pastebin of the issue but it is posting the same address to my two services [18:57] hmm now im getting this ERROR environment "manual" has an unknown provider type "manual" [18:59] Valduare: are you on 1.16 ? [18:59] ya ran brew from my mac [18:59] Valduare: then the provider type "provider:" is null [18:59] 1.17 changes that to manual [18:59] so you can name the parent key null [18:59] but the provider key needs to be null [18:59] ah [19:00] null or “null:” [19:00] Valduare: I think you can brew install the development version, 1.17 [19:00] if you're interested [19:01] i am [19:03] there we go - got it installing 1.17.3 [19:03] eek, that needs to be updated [19:03] * marcoceppi opens a request to update brew [19:03] 1.17.7 is latest devel === Ursinha is now known as Ursinha-afk [19:09] woo its doing something heh [19:09] looks like its connected to bootstrap-host [19:09] ah it failed [19:09] ERROR Bootstrap failed, and the environment could not be destroyed: rc: 1 [19:17] marcoceppi: looks like im out of luck on mac? [19:17] till brew gets 1.17.7 [19:17] the bug lists it fixed in that version [19:18] Valduare: I'll open a merge requset in second === Ursinha-afk is now known as Ursinha [19:36] marcoceppi: ty - how long will it take to update there you think? [19:36] depends on whenever they accept the pull request [19:39] ok [19:39] can you install juju on more than one workstation to manage a set of machines? [19:40] Valduare: yes [19:40] ok i’ll reboot into ubuntu real quick brb [19:40] ty for this help [19:52] hi [19:52] ok on ubuntu [19:52] installed the ppa and cant find 1.17.7 [19:53] Valduare: ppa:juju/devel [19:54] oh ... ya lol [19:54] need more coffee [19:55] can you guys bootstrap and deploy with 1.17.7 locally on lxc? [19:55] mine failed to deploy, it tried to download 1.16.6.1 tools === CyberJacob|Away is now known as CyberJacob [20:04] marcoceppi: ok got it setup 1.17.7 [20:04] do I need to run commands on my local workstation as sudo ie sudo juju bootstrap or just juju bootstrap [20:05] you shouldn't ever run sudo with bootstrap if you're using juju 1.17+ [20:05] ok [20:05] says the machine is already provisioned due to that failed one from my mac [20:05] how do I start it fresh [20:07] Valduare: ssh in to the machine [20:08] ah it worked [20:08] ran the command again [20:10] so now thats done, whats next [20:10] im at bottom of this guide now https://juju.ubuntu.com/docs/config-manual.html [20:17] Valduare: you have a machine registered in the manual provider [20:17] you're free to deploy services now [20:17] eg: juju deploy mysql --to 1 [20:18] nice [20:18] so this first machine the bootstrap can be used for anything I want? or is it used to manage other machines? [20:19] its the state server, I wouldn't put anything more than the JujuGUI on it [20:19] lazyPower: heya.. quick question.. with ansible, one command will completely finish before the next one starts, correct? [20:19] * hatch perks up [20:19] someone said juju gui :) [20:20] cjohnston: correct [20:20] hrm [20:20] ah ok I got juju gui on it now [20:20] cjohnston: its going to wait on return codes before it procedes. If a command exits > 0 it will halt execution unless you tell it to ignore failure. [20:21] ansible is a FIFO style exection stack, none of it is async sadly. [20:21] lazyPower: I'm trying to do an unarchive, it keeps erroring saying the file isn't there, however, by the time I'm able to login and look, it's there [20:22] oh actually i stand correcte [20:22] *corrected [20:22] you can poll for the file presence [20:22] http://docs.ansible.com/playbooks_async.html [20:23] ok so when adding new machines with manual provider, all my vm's are behind NAT so I will needto specify a ssh port for juju to connect to them? [20:23] cjohnston: wait again, thats not going to help. Thats about running against multiple nodes. [20:23] :( [20:24] but apparently it works when embedded on a task. Verified locally by waiting for a directory to be created [20:25] https://gist.github.com/chuckbutler/9842221 [20:25] ack, I'll give it a try [20:25] ta [20:28] cjohnston: so, you could also use the wait_for module. [20:28] - wait_for: path=/tmp/beepdoop [20:28] it will hang until it shows up, without that fugly bash [20:29] is there somewhere that lists system requirements for charms [20:29] ie suggested ram and such for the wordpress charm [20:29] Valduare: if it has constraints it should be listed in the readme. [20:29] ok [20:29] otherwise, all charms assume 1g of ram and 1 cpu [20:29] and what about the ssh port question [20:30] Valduare: ah i haven't provisioned behind a nat, so i dont know. [20:30] maybe I should be doing it a different way? [20:31] you're communicating with an ESXI host right? [20:32] k I have a freenas box for storage with iscsi to an esxi host, on the esxi host I run a smoothwall virtual router with two nic's, one nick for publicip (red) and one nic for local (green) then I NAT ports to the local ip of the server that hosts such and such app [20:33] all running on an amd e350 mini itx board and 4 old sata 1.5 400 gig drives in zfs mirrors / striped together [20:33] hmm... i was going to say bridged networking may be the key [20:33] but sounds like you're already pretty much there [20:34] aye only 5 public ip's without going crazy on monthly bills :) [20:34] yeah, with that virtual router sitting in the middle, thats going to be your meal ticket. [20:34] so hence the smoothwall virt routers taking up a public each and spliting off [20:35] so pretty much treat each virt router like a bank of servers [20:35] so i'd have a juju bootstrap under each virt router? [20:35] no, you can manage all of them from a single juju bootstrap [20:35] you'll have to register the connection with one on a non standard port and see what happens [20:36] otherwise you're burning resources with a state controller under each vnet. [20:36] maybe setup another virtual network connecting them? [20:36] * lazyPower shrugs [20:36] give it a go and let me know what happens [20:36] i'm curious about the result [20:36] im suprised what i've bene able to accomplish so far with the hardware I have :P [20:38] im scrolling back up - but what did we figure out was the way for juju to contact new machines [20:39] ie since I have to specify an ssh port other than 22 === CyberJacob is now known as CyberJacob|Away === CyberJacob|Away is now known as CyberJacob [21:02] hi all, I am trying to bootstrap a local environment on trusty but I am getting the following: "ERROR no matching tools available". [21:02] trusty is not final yet, so should I download this locally from somewhere else? [21:09] note that I am using juju client for OSX against a local trusty vagrant box [21:13] hmm cant access juju gui after changing NAT for port 8080 to port 80 locally [21:21] Valduare: so, i just setup maas [21:21] yah? [21:21] Valduare: i rescind what I said. Its a pain in teh neck to setup, but this is pretty awesome [21:21] :) [21:21] i dont think the hassle is worth it for less than say, 10 machines though [21:22] if you're only working wiht 5 or 6, manual is a far lower barrier to entry [21:22] I got 14 or so right now [21:22] im stuck right now- how do I add a second host to juju on a nonstandard ssh port [21:26] user@host:port [21:26] Valduare: its user@host:port [21:26] ninja'd [21:26] tried that [21:26] says already provisioned [21:26] but its not. [21:27] wonder if I can config my ssh config and go that route? [21:27] Valduare: welll, that means that you've tried to provision at one point [21:27] Valduare: you'll need to ssh in to that machine [21:27] or. its trying on port 22 wich is machine 0 [21:27] run `sudo initctl list | grep juju` [21:30] looks like its working after setting up ssh config [21:30] it was disreguarding ports with the juju command for some reason [21:32] how can I change the port juju gui is running on [21:39] im at a loss as to how to have a web server and juju gui running at same time lol === hatch__ is now known as hatch [21:53] lazyPower: ansible appears to be ignoring my wait_for :-( [21:54] cjohnston: that, doesn't sound fun [21:57] lazyPower: any ideas: http://bazaar.launchpad.net/~cjohnston/+junk/ans-graphite/view/head:/graphite.yaml === CyberJacob is now known as CyberJacob|Away [22:12] cjohnston: not off the top of my head [22:12] cjohnston: have you asked in #ansible? [22:17] no, will do [22:30] on juju gui how do you specify --to [22:34] Valduare you can't....yet [22:35] darn [22:35] we are working hard on that [22:35] so just view only interface for me then :P [22:35] Valduare.....soon....ish :) I promise [22:35] now im trying to wrap my head around how the wordpress setups go with the charms [22:36] ie i understand how i'd setup a single site [22:36] but how can I setup a single vm with multiple sites if I am hardware constrained [22:36] if that's not in the documentation your best bet would be to get in touch with the author [22:38] ie when I deployed wordpress, it just put wordpress in /var/www/ not within a dir of its own - which it should be for future expandability etc [22:40] You may want to ask http://askubuntu.com/questions/ask?tags=juju,wordpress [22:41] or file a bug https://bugs.launchpad.net/charms/+source/wordpress if that bears no fruit [22:45] ok [22:46] this is interesting stuff :) === Ursinha is now known as Ursinha-afk [22:53] single VM with multiple sites = candidate for LXC? [22:53] each would see his own file system [22:53] so no problems to put each charm in the smae place === Ursinha-afk is now known as Ursinha === arosales_ is now known as arosales [23:55] hi guys