=== gary_poster is now known as gary_poster|away === freeflying_away is now known as freeflying [00:35] I released a node from maas, which has been deployed with service by juju, is it possible to destroy the service now [00:42] freeflying: cn you say what you did another way [00:42] i am not a maas expert [00:44] davecheney, I have a environment running, deployed ceph charms, it fails due to lack of a second hd, then some one reinstalled aubuntu on one of the machine, I tried to destroy -service, [00:45] davecheney, I can't achieve it, use destroy-unit -> resolved -> destroy-machine, all turned to fail too, then I release the node from maas, but juju status still hows its there [00:45] freeflying: yes, sorry this is a known issue [00:46] davecheney, any workaround so far [00:47] https://bugs.launchpad.net/juju-core/+bug/1206532 [00:47] <_mup_> Bug #1206532: --terminate option for destroy-unit [00:47] freeflying: apart from ignoring the entry in status [00:47] no [00:48] https://bugs.launchpad.net/juju-core/+bug/1089289 [00:48] <_mup_> Bug #1089289: remove-unit --force

[00:48] and several others [00:48] davecheney, can we delete the status from mongodb [00:49] freeflying: possibly, but utterly not recommended === defunctzombie is now known as defunctzombie_zz === scuttlemonkey_ is now known as scuttlemonkey === freeflying is now known as freeflying_away [04:03] having problems getting juju bootstrap to work on a fresh ubuntu instance [04:03] not sure if i have my config set correctly. [04:04] and for whatever reasons, juju -e tosses an error too [04:08] error: error parsing environment "hpcloud": no public ssh keys found [04:09] what am I missing? [04:14] authorized-keys-path is missing from the default config. [04:14] appears to work (well returns nothing) after I added that === freeflying_away is now known as freeflying === defunctzombie_zz is now known as defunctzombie === thumper is now known as thumper-afk === CyberJacob|Away is now known as CyberJacob [06:38] juju credential error masage | http://askubuntu.com/q/350187 === freeflying is now known as freeflying_away === defunctzombie is now known as defunctzombie_zz [09:26] Does juju helps to build own private cloud | http://askubuntu.com/q/350226 === thumper-afk is now known as thumper === CyberJacob is now known as CyberJacob|Away === gary_poster|away is now known as gary_poster === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away [12:58] Problem with upload images openstack | http://askubuntu.com/q/350243 [13:03] mrz, did you get your hp config working? [13:05] arosales: thought you'd tyoped me then... [13:06] mgz, ha I also initially thought you were also asking an hp question, and I thought that couldn't be right :-) [14:20] orangesquad: can i get a review of https://code.launchpad.net/~jcsackett/charmworld/askubuntu-job/+merge/187758 and https://code.launchpad.net/~jcsackett/charmworld/askubuntu-in-review-queue/+merge/187771 please? neither are very long. [14:21] jcsackett, I am in meetings for the next 3 hours. I can look after then [14:27] hi all [14:27] hey utlemming [14:28] where are you come from? [14:37] <_mup_> Bug #1229275 was filed: juju destroy-environment also destroys nodes that are not controlled by juju

[14:52] hey epafrashg_ anything we can help you with? [15:48] jamespage: sinzui: arosales: hah! 12.04.3 fresh install ... did the updates, installed the PPA, apt-get install juju-core juju-local ... [15:48] and ... everything is working [15:49] \o/ [15:53] arosales: more or less. i ran into a service group limitiation [15:53] mrz, ah yes that has bitten me too [15:53] the osx brew keeps bailing on me so i gave up and spun up an ubuntu instance [15:53] an HP ticket should resolve that [15:54] i have to fill out a form to get it increased [15:54] mrz, funn you mention that I was just talking to marcoceppi and sinzui about the juju osx client [15:54] mrz I have asked HP support to just give a general bump the the sec group limit [15:54] 10's small [15:55] mrz, agreed. They really didn't give me an answer. They are evaluating it for bumps across the board [15:55] i haven't played with it much but how does scaling work? is thats some hook wihtin the charm that shoves things behind ha-proxy or something? [15:55] i have access to the East coast beta but that's no better [15:56] add-unit spins up another instance and mongo keeps track of it in relation to the service [15:56] https://juju.ubuntu.com/docs/charms-scaling.html [15:56] but it doesn't go into the "how" [15:57] mrz https://juju.ubuntu.com/resources/overview/ goes into the "how" Under "Scaling services horizontally" [15:57] the example walked through adding 100 instances of mysql but it's really only magic if those are read-only slaves behind a load balancer [15:58] hazmat, marcoceppi what is the command to disable sec groups? [15:58] oh cool, i'll read that today [15:58] mrz, I think the mysql charm itself can set up a cluster with slaves [15:58] and you can add-unit behind that [15:59] but it read like it was just master/slave and not some sort of load balancing between slaves [16:00] arosales, disable sec groups? [16:00] arosales, oh.. firewall global mode [16:00] hazmat, thats it [16:00] hazmat, ya to get around hp sec group limits [16:00] its an undocumented config... evilnickveitch it would be good to capture that one [16:00] atlast, Juju for Azure :D [16:00] hazmat: yes, for cases where all my stuff is behind a load balancer or just internal. [16:00] hazmat: firewall-mode: global [16:01] mrz, put that into your HP stanza in environments.yaml _if_ and only if you want to side step security [16:01] ie doing some dev [16:01] arosales, we didn't need to use it for the training.. hp up'd our limits, and we split the class across two zones. i had a nightly setup to clear out the unused groups and env resets. [16:01] hazmat, thanks [16:02] arosales, I spoke to the HP people about that, oh gosh, about 3 months ago, but they never got back to me either [16:02] evilnickveitch, ya . . [16:02] * arosales sighs [16:03] hazmat, thanks [16:03] hazmat, good to hear for the training, but a nice option to have in your back pocket for development education purposes [16:03] arosales, absolutely [16:07] * arosales strongly notes for education and development purposes :-) [16:07] arosales: https://bugs.launchpad.net/juju-core/+bug/1229903/comments/3 [16:07] <_mup_> Bug #1229903: Default to local provider [16:08] "firewall-mode: global" does what exactly? [16:08] does it use a default group? [16:08] (it's sort of a pita to add security groups after instance creation right?) [16:08] jcastro, for completeness could you take a 12.04.0 and install saucy lxc on it and see how local provider works? [16:09] arosales: but I'd need the kernel too? [16:09] mrz: I think it disables the use a per service sec group [16:09] arosales: or do you mean, install .0 and then do what I recommend in the bug to confirm? [16:09] arosales: i like that. wonder what it uses then. [16:10] jcastro, the thought is with and updated lxc you wouldn't need an updated kernel [16:10] huh [16:10] ok [16:10] jcastro, thanks [16:11] mrz I haven't used it personally, but I am _guessing_ it reuses [16:11] marcoceppi, or hazmat may have had more actual experience. [16:11] arosales: suppose i should just try it [16:12] i don't think i've had luck using the nova client to adjust groups post instance creation [16:12] mrz, it uses a single group and adds per machine entries to it [16:12] hp support does get back pretty fast for these types of requests [16:12] arosales: we'd still want people on the newer kernel anyway though right? [16:12] but if you wanted to try for development purposes may be worth a shot [16:12] I mean, if you're going to use LXC we want them on the newest kernel we can support right? [16:12] mrz default juju would use a single group per machine [16:12] hazmat: oh. [16:12] jcastro, not necessarily [16:13] mrz, part of this is throw back to ec2 where security group to machine is static at machine creation outside of vpc [16:13] 12.04.0 is supported and gets updates [16:13] openstack is a bit more flexible.. and can do runtime mapping of groups to machines [16:13] jcastro, for a good user story it would be nice to not have to reboot [16:13] * jcastro nods [16:14] but the latest kernel isn't that bad, but its an unknown what a user has tied to a kernel. I think testing and updated lxc on precise may just give us some more information to recommend. [16:15] * jcastro nods - I should have results for you in about ~30 [16:15] "Scaling services horizontally [16:15] oops [16:15] jcastro, thanks [16:15] "Scaling services horizontally" - doesn't really tie into the next step of getting both "web servers" behind a load balancer. [16:15] i suppose that'd be another relationship to build? === defunctzombie_zz is now known as defunctzombie [16:15] yes [16:15] you do ... [16:16] juju deploy haproxy [16:16] then `juju add-relation haproxy whateveryourservice` [16:16] then you `juju unexpose whateveryourservice` [16:16] and then `juju expose haproxy` [16:16] right, that totally makes sense then. [16:16] then `juju status haproxy` to get the public IP, then update DNS [16:16] however ... [16:16] some charms have loadbalancing in them [16:17] so like if you `juju add-unit wordpress` each head registers itself with the nginx load balancer on all the other heads [16:17] so you don't need an haproxy there [16:18] ideally a charm would have options for either way of doing it === defunctzombie is now known as defunctzombie_zz [16:20] mrz: hmm, maybe I should add the haproxy example to that page? [16:20] ah. i wasn't aware the wordpress charm worked like that. so in the case of wordpress, how do I figure out which IP to add to dns? [16:20] any one of them [16:21] but then i'd need to make sure i don't destroy that instance right? [16:21] the nodes just figure it out, it's kind of badass. [16:21] yeah [16:21] i wonder how youi'd safe guard againts humans doing Bad Things there. [16:22] jcastro: and yeah, the ha-proxy example would be good. [16:22] you know i should hack up wordpress to use HP's cloud storage vs. nfs. [16:23] scares me to think of running nfs in the cloud like that [16:23] half off pricing! [16:24] it's what I liked about azure (minus the pricing part). the WP Azure plugin rocked it. [16:24] i have no idea where themes ended up but largely i didn't need to care. [16:29] marcoceppi: hey, does the mediawiki charm have the same load balancing set up as wordpress? [16:29] jcastro: nope [16:29] from looking at it I don't think it does [16:29] hah, so the docs are totally incorrect [16:32] jcastro, erm, well, not totally. s/mediawiki/wordpress would fix them. [16:32] though it would be nice to add some hapraxy eaxample instead [16:34] I am doing both now [16:34] expect an MP in a few === defunctzombie_zz is now known as defunctzombie [16:43] evilnickveitch: https://code.launchpad.net/~jorge/juju-core/scaling-fixes/+merge/187847 [16:44] cool. i just need to make a few tweaks === defunctzombie is now known as defunctzombie_zz [16:45] mrz: thanks for that feedback, that was really useful! [16:45] np. [16:48] marcoceppi: evilnickveitch: I suspect the original author mixed up wordpress and mediawiki charms [16:48] which is like the thousanth time that's happened [16:49] jcastro, possibly, well done for spotting it though, that page has been through umpty-ump revisions and nobody noticed before [16:51] I guess because it doesn't generate an error, just doesn't work as intended [17:09] jcastro, merged [17:17] marcoceppi, sinzui: got brew to install juju [17:17] http://pastebin.ubuntu.com/6159550/ [17:18] marcoceppi, sinzui: I had to have xcode on 10.8 installed, and accept the license [17:18] in addition to the command line tools the brew wanted me to install [17:22] arosales: ok so testing with just "lxc" isn't working out [17:22] the thing is to bring it back it has deps on other things in raring [17:22] so it wants to pull in a new libc6, etc [17:28] jcastro, I was worried about the deps :-/ [17:28] jcastro, is that dep string pretty long? [17:29] 23 deps [17:29] marcoceppi, sinzui: I am not sure if that means for https://github.com/mxcl/homebrew/pull/22772 and 1.14.1 if xcode is a dep. I couldn't get it working with out xcode. I'll post a comment [17:29] jcastro, ouch [17:29] including some libapparmor stuff and libnih stuff that looks scary [17:29] hwe may be the way then [17:29] jcastro, thanks for investigating [17:29] jcastro, can you also add that comment to the bug? [17:30] but the new libc6 would mean they'd have to reboot anyway, so this is actually worse than the HWE experience [17:30] yeah [17:37] jamespage: will the raring version of python-quantumclient work on precise? [17:39] jcastro: I have openstack/juju/MAAS working fully on VMWare Fusion and Workstation now. [17:39] next step is building redundancy in === defunctzombie_zz is now known as defunctzombie === CyberJacob|Away is now known as CyberJacob === scuttlemonkey_ is now known as scuttlemonkey === defunctzombie is now known as defunctzombie_zz === defunctzombie_zz is now known as defunctzombie === defunctzombie is now known as defunctzombie_zz === defunctzombie_zz is now known as defunctzombie === defunctzombie is now known as defunctzombie_zz [19:29] benji, gary_poster gmail + juju-gui list == war. They won't let me reply [19:29] benji, gary_poster: http://pastebin.ubuntu.com/6160045/ [19:30] thank you very much Curtis! I'll paste that into the conversation so others can read [19:31] I am still getting re-subscribed under an address that the two systems accept. [19:32] sinzui, meanwhile I told the mailing list to accept that other email forevermore. [19:32] Thank you! [20:22] Switching Juju lxc bridge | http://askubuntu.com/q/350503 === defunctzombie_zz is now known as defunctzombie === defunctzombie is now known as defunctzombie_zz === CyberJacob is now known as CyberJacob|Away === sidnei` is now known as sidnei === defunctzombie_zz is now known as defunctzombie [22:21] jcastro: ping === defunctzombie is now known as defunctzombie_zz === defunctzombie_zz is now known as defunctzombie