[02:37] lazyPower: hye! are u available? [02:37] lazyPower: I did follow this tutorial but I failed to restart my networking service [02:39] lazyPower: then I reboot, and affect to show my juju status [03:45] hye, does anyone can help me , what should I do if my juju status mentioned agent-state-info : container failed to start? [03:50] hye, does anyone can help me , what should I do if my juju status mentioned agent-state-info : container failed to start? I'm install juju using lxc in local provider [04:15] halcyon: this is more than likely related to reconfiguring your LXC bridge according to that post. As i mentioned - its tricky [04:15] and failure is difficult to debug when that happens [04:15] halcyon: how much linux networking do you know? [04:16] or rather, how confident are you in the event of catastrophic failure that you could reconfigure it into a working condition :) [04:18] lazyPower: just a beginner :) [04:19] lazyPower: could u advice me on what should I do? [05:48] halcyon: Revert back to the LXCBR0 bridge, and revert the networking + subnet back to the 10.0.3.0/24 cidr [05:49] halcyon: i'm coming in and out of sleep - so if you're around in EST morning hours i can help. but for now i gotta get some sleep. have a good evening o/ [08:33] good morning o/ [08:35] I'm having problem with my first charm [08:37] 1 - when I do a juju-log in the bash code of the charm, where can I actually find those logs? [08:37] because I have a "hook failed: install" but can't figure why [08:41] maybe I'm not using well the "set -e" thing... [08:42] Muntaner, in the unit log [08:43] Muntaner, e.g. if you have unit wordpress/2 on machine 4, juju ssh 4, sudo less /var/log/juju/unit-wordpress-2.log [08:43] dimitern, figured it [08:43] seems like I had an error caused by a "mysql restart", and I didn't have mysql on that machine [08:43] (dumb errors :) ) [08:44] Muntaner, yeah - hooks need to be idempotent - calling them multiple times shouldn't be a problem [08:46] dimitern, nice! it works :) [08:46] now, I need to connect the mysql charm to my charm [08:46] can you tell me how to do that? no ideas, right now [08:48] Muntaner, you'll need to specify in your charm metadata that you require a db relation like wordpress does for example [08:48] Muntaner, then you can add-relation yourcharm mysql [08:49] Muntaner, and ofc the relevant relation hooks need to be in place (db-relation-changed, etc.) [08:54] dimitern, maybe I'm getting it to work :D [08:54] Muntaner, cool :) [08:54] dimitern, I need to tell to my charm the IP of the mysql charm... can I do that? [08:57] Muntaner, check out https://jujucharms.com/docs/authors-charm-interfaces [08:58] Muntaner, and this https://jujucharms.com/docs/interface-mysql [08:58] dimitern, ty! [09:12] Hello, guys. I have a working juju environment, inside OpenStack. I have my bootstrap node and various other nodes with services happily running. I just managed to destroy my ~/.juju directory, along with the .jenv file that's tied to my environment. [09:13] Is there a way to extract those info, somehow, from the bootstrap node? [09:22] dimitern, seems like I'm having problems in running the "db-relation-x" hooks [09:23] logs aren't helping me [09:28] dimitern, sorry I'm in a meeting perhaps lazyPower or some of the charmers can help you? [09:31] dimitern, no problem! === scuttlemonkey is now known as scuttle|afk [10:07] dimitern, my charm 0.1 is ready [10:08] Muntaner, awesome! does it work with mysql? [10:08] dimitern, yep! [10:08] prestashop 0.1 ready ;D [10:09] I should implement other hooks now [10:09] Muntaner, sweet! [10:11] dimitern, how can I provide to the guy that deploys via juju-gui or CLI some informations about the default log-in, the default link to reach to see the web service, etc. ? [10:12] Muntaner, I guess in the readme file for the charm [10:24] dimitern, can't access to readmes via the gui - seems like it loads forever, strange [10:27] frankban, hey, perhaps you can help ^^ [10:30] Muntaner: so you are not able to see the readme files from the GUI? for all the charms? [10:30] yep frankban, first time that this happens [10:30] I see "Loading..." and the spinning forever [10:31] Muntaner: ok I am trying to dupe [10:32] frankban, tell me if you need anything (logs, background) [10:32] Muntaner: sure thank you [10:33] frankban, I am logged in the same juju-gui endpoint with two laptops, if this may help [10:33] same thing happens with only 1 guy logged, just tried [10:35] on 1.20.14, on openstack provider, is there a way for me to get hold of the private ip address of a unit that has been assigned a floating ip? [10:35] juju status just shows the public ip, not private [10:38] frankban, another little issue: I can see the icon of my charm in the canvas, but on the left side I see its default icon [10:39] Muntaner: I just deployed the GUI on a ocal env using quickstart, and when I click charms on the left I can see their readme, configuration, etc without problems [10:40] frankban, maybe I caused this error someway [10:40] Muntaner: could you please file a bug describing the steps to dupe the error? [10:43] frankban, I just created my personal charm, changed his icon, deployed it via CLI [10:45] Muntaner: deployed as a local charm? [10:45] frankban, I have an OpenStack private installation, all-in-one. The charm is a prestashop deployer - deployed it via CLI in the same LAN of the OpenStack server [10:46] Muntaner: what command did you use to deploy the charm? [10:46] frankban, juju deploy --repository=/home/mike/charms local:trusty/prestashop --debug [10:47] Muntaner: ok so this is a local charm deployment [10:47] Muntaner: how do you use the GUI to try to read the readme? [10:49] frankban, I click on the charm on the canvas, then I click on the link on the left tab (for example: cs:trusty/mysql-22), and then it opens a tab that loads forever [10:49] if you need screenshots, can up them [10:50] Muntaner: no thanks, now trying to dupe what you described [10:55] Muntaner: ok I deployed a local charm and now I see the readme error [10:56] great frankban! [10:57] Muntaner: oh... after refreshing the GUI, I can now see the readme of the local charm, could you try it? [10:59] frankban, F5ed a couple of times - no changes [11:00] Muntaner: any JS console errors? [11:01] Muntaner: oh I think I know why, I guess your charm is not in the charm stor eyet, correct? [11:01] no frankban, it justs exists in my laptop folder [11:02] Muntaner: ok === Murali_ is now known as Murali [11:16] frankban, btw I can't read the readme of any charm (juju-gui, mysql) [11:16] Muntaner: even after refreshing? [11:16] frankban, yes [11:17] Muntaner: could you please try to refresh removing the URL path? so just the home of the GUI [11:19] Muntaner: anyway, I filed https://bugs.launchpad.net/juju-gui/+bug/1427162 , please feel free to add there any additional information you have. and thanks for reporting this issue! [11:19] Bug #1427162: Local charms icon and readme === Murali_ is now known as Murali [11:54] gnuoy, can you do me a quick review of https://code.launchpad.net/~james-page/charm-helpers/bug-1413862/+merge/251426 [11:55] sure [11:57] jamespage, approved [11:57] gnuoy, ta === Murali_ is now known as Murali === scuttle|afk is now known as scuttlemonkey [14:25] hey guys [14:25] http://askubuntu.com/questions/590715/how-can-i-find-power-pc-specific-charms [14:25] we don't have this do we? [15:25] jcastro: nope.. tags might help for the short term [15:28] jcastro: yea, there was talk of adding a metadata.yaml entry for that but nothing was firmed up and planned that I'm aware of. [15:29] rick_h_, I was thinking though, if we see which charms fail on ppc64le we could have a general idea? [15:31] jcastro: maybe, I need to catch up on where/how good charm testing is to be a good indicator of something like this [15:36] rick_h_, want me to answer it to that effect? [15:37] hey tvansteenburgh ^^^ [15:39] jcastro: we're not testing on ppc64le yet [15:39] not in CI anyway [15:39] ack [15:49] http://askubuntu.com/questions/590715/how-can-i-find-power-pc-specific-charms/591878#591878 [15:49] How's this guys? [15:50] mbruzek, looks like that guy's jruby fixes for ppc64le are in upstream now [15:50] jcastro that is great [15:52] hey guys [15:52] just a quick question [15:52] how can I get in a juju hook [15:52] for example, install or db-relation joined [15:52] what do you mean by in a juju hook [15:52] "get in"? [15:52] the floating IP assigned to the VM that is running this charm? [15:52] sorry, bad english ^^" [15:53] https://jujucharms.com/docs/authors-hook-debug [15:53] ok so this shows you how to go into a unit and debug a hook [15:53] I assume your install hook failed and you want to see why? [15:53] Muntaner: the public-address and private-address is always available to a hook, likewise those values are available on relation hooks for the other charms. [15:54] no jcastro, I am running a command line installation into the install hook. I need to specify the local ip of the VM as a parameter of the installation [15:55] oh yes, then what mbruzek said, let me find the doc snippet for those [15:55] Muntaner: if you want the private address for the system you are installing use 'unit-get private-address' [15:56] Muntaner: https://jujucharms.com/docs/authors-hook-environment [15:57] mbruzek, thanks, lemme try [15:57] Muntaner: if you need the address of a related charm, say in the db-relation-changed hook, use `relation-get private-address` [15:58] evilnickveitch, hey since we're ditching the charm feature rating bits, can I go ahead and snip out that page and submit a PR? [15:58] jcastro, if we are doing that then sure [16:00] ok confirmed we are, working it now [16:13] mbruzek, echos tell me that it's working, thanks :) [16:13] Muntaner: you are most welcome [16:20] mbruzek, [16:20] now I'm having a very strange issue [16:23] seems like the commands relation-get database return nothing [16:23] and they always worked [16:24] Muntaner: Yes I have run into that before. Sometimes the hook is fired before all the data is available. Many database relation hooks exit 0 (silently) until the database' [16:24] key is in the relation-get [16:25] Muntaner: which is the reason all hooks must be idempotent they could be called several times. An operation like building a database takes a little bit of time, so it is possible to have the relation fired before it is done. [16:26] mbruzek, the strange thing is that if I put the unit-get etc I have the problem [16:26] if I comment it, the problem isn't present anymore [16:29] Muntaner: Check out the following code I wrote: https://api.jujucharms.com/v4/precise/openmrs-1/archive/hooks/database-relation-changed [16:29] I followed the database-relation-changed pattern. Get all the database values and in `relation-get database` did not return anything `exit 0` because the relation could be called several times. [16:31] mbruzek, sorry, I can't get it [16:31] I'm not understanding the "problem", if there is one :( [16:38] mbruzek, sooo... how can I be sure that all the values aren't empty? [16:39] Muntaner: check them all? [16:40] The code checks the "database" variable because that is the last one to be set. If that one is available the others should be too. [16:44] mbruzek, if I do an exit 0... do the hook "finish" ? [16:45] If you exit 0 it stops there without an error [16:45] If you exit 1 Juju reads that return code as an error. [16:45] Muntaner: The hook will be called again when database is set [16:46] mbruzek, it seems like my hook did an exit 0 [16:46] and in fact, his execution stopped in half [16:48] mbruzek, it does not seem to be called again... so my charm is in an "half deployed" status [16:50] Muntaner: What are you writing the charm in? Use juju-log or echo or print() to debug this. If the charm is half deployed I would think you need to exit earlier if you don't have the data you need. Only do the configuration when you have all the data [16:51] mbruzek, it's simple linux shell bash [16:52] always use set -ex for bash scripts so you can see the individual commands running. [16:53] mbruzek, already using it [16:53] Muntaner: then you should be able to figure out why the charm is half configured. [16:54] yes mbruzek, I know where I call the exit 0 because one variable is empty [16:54] but I can't figure how should I implement a workaround for this [16:54] Muntaner: exit 0 early if you don't have all the information, when you have all the data continue to set up the whole configuration. [16:55] mbruzek, I get lost at the "when you have all the data" part [16:55] All the database variables, and the private address, what ever your charm needs [16:56] yer mbruzek but I can't understand how can I check "when" I have them, since the hook exited [16:56] Muntaner: the hook will be called again when you get more information [16:57] mbruzek, btw, I'm not launching mysql and mycharm simultaneously: mysql is started since a long time [16:57] it's the same situation? [16:57] (I mean the mysql charm from the store) [17:01] Muntaner: Yes this is a common patter to connect to a database charm. You can launch them simultaneously, the db-relation-changed hook can be called multiple times until you get all the database variables [17:04] mbruzek, maybe I got it. my relation-changed is empty, actually I'm only writing in the relation-joined [17:04] Muntaner: well there is your problem. [17:30] mbruzek, now I echo some dumb stuff in the relation-changed and seems to work [17:38] bye guys, see you tomorrow :) [17:38] mbruzek, thanks for the tips [17:38] Muntaner: you are welcome [17:38] Muntaner: glad you got it working. [19:22] Hey marcoceppi. If I wanted to rsync files between charms how would I set up creds? [19:22] mbruzek: is this a peer relation or a service relation? [19:22] suddenly wget doesn't seem so horrible ;) [19:22] marcoceppi: yes, that is the hook that I would use. [19:23] peer? [19:23] no this would be a service relation [19:23] icky, you'd have to have some kind of key exchange on the interface [19:23] which isn't worth it anymore imo [19:23] I thought this was a peer relation [19:23] How does peer relation make it better? [19:24] mbruzek: the interface is virtually irrelevant, you can add any keys you want because it's an implicit relation between the units in the service/charm and not to other charms [19:25] I see. [19:26] marcoceppi: do you know of any charms that transfer *anything* via rsync ? [19:27] i know they exist, not off the top of my head, but they're almost always peer relations [19:27] This would be more of a master/minion relation [19:28] Over the banana interface [19:28] mbruzek: i thought we refactored that to the starfruit interface [19:28] so this is an interface you're creating? [19:28] marcoceppi: yes [19:28] marcoceppi: I have complete control [19:28] mbruzek: then make the public key of that machine published on the interface [19:29] public-key=`cat ~/.ssh/id_rsa.pub` [19:29] and have all the units that get that add that public key to their authorized-keys file [19:29] then you can rsync to all the units [19:30] http === kadams54 is now known as kadams54-away [19:50] or, http === anthonyf is now known as Guest30629 === kadams54-away is now known as kadams54 === Guest30629 is now known as anthonyjf === urulama is now known as urulama|afk === roadmr is now known as roadmr_afk === kadams54 is now known as kadams54-away === urulama|afk is now known as urulama [20:32] whit, link me up! [20:33] for jujubox? [20:33] jcastro, one sec [20:33] jcastro, https://registry.hub.docker.com/u/whitmo/jujubox/ [20:34] jcastro, if you use docker 1.5.0 you can build and run charmbox, the second docker file [20:35] jcastro, jujubox is only good for deploying, you need charmbox if you are doing any serious reviewing [20:35] ack [20:35] hey so, right off the bat [20:36] in the "Once it's up" section in your docs [20:36] put quickstart first, because you can't status without bootstrapping [20:37] whit, also ran into this: http://pastebin.ubuntu.com/10506159/ [20:37] whit, ah I see, perhaps mention up front that the local provider won't work [20:41] whit, have you posted this to the list? I don't see it there [20:42] yeah that's mentioned in the "future" section but we should put it more prominently [20:42] jcastro, it's only been mentioned as an aside wrt review queue [20:43] jcastro, if you mount your $JUJU_HOME, status will work if you have an env bs'ed, but yeah will fix that for the barecase [20:43] nod [20:44] I was thinking more for the from-scratch case. [20:44] jcastro, exactly [20:44] jcastro, hypothetically it should run anywhere docker does [20:45] jcastro, which is a nice story for folks who are not on ubuntu but want to try juju [20:48] whit, man this is fast [20:48] much faster than the vagrant box, heh [20:48] jcastro, yup === kadams54-away is now known as kadams54 === roadmr_afk is now known as roadmr === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [21:48] Juju peeps. My LXC environment seems to be corrupted. agent-state-info: 'failed to retrieve the template to clone: error executing "lxc-create": [21:49] Does anyone know how to fix something like that? I tried cleaning out the LXC images, but no go [21:50] $ sudo lxc-destroy -n mbruzek-local-machine-2 [21:50] lxc_container: lxccontainer.c: container_destroy: 2041 Error destroying rootfs for mbruzek-local-machine-2 [21:50] help? [21:50] mbruzek: thumper is your local provider guy :) [21:51] mbruzek: what does `sudo lxc-ls --fancy` show? [21:54] waigani: pastebin.ubuntu.com/10506962/ [21:55] waigani: I tried to destroy the stopped containers but I can not. [21:56] waigani 1.21.3-vivid-amd64 if that makes any difference [21:56] mbruzek: have you tried with -f [21:56] yes [21:56] had to ask :) [21:58] $ sudo lxc-destroy -f -n mbruzek-local-machine-3 [21:58] lxc_container: lxccontainer.c: container_destroy: 2041 Error destroying rootfs for mbruzek-local-machine-3 [21:58] Destroying mbruzek-local-machine-3 failed [21:58] mbruzek: lxc-info -n mbruzek-local-machine-3 [21:59] Name: mbruzek-local-machine-3 [21:59] State: STOPPED [22:00] waigani http://pastebin.ubuntu.com/10507040/ === kadams54 is now known as kadams54-away [22:08] https://bugs.launchpad.net/juju-core/+bug/1307215 [22:08] Bug #1307215: destroy-environment fails to clear lxc containers

[22:09] I used the work around deleting /var/lib/lxc/ [22:11] mbruzek: sorry interrupted, I was just about to send you that link. Beat me to it. === kadams54-away is now known as kadams54 [22:17] Thanks for your help waigani [22:18] mbruzek: np === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr [22:33] hey cory_fu finally got back to your review comments. [22:33] corey_fu not sure I understand the .git inclusion one [22:37] whit: Yeah, I replied to your comment on that one. [22:37] cory_fu, ah thanks [22:38] whit: Regarding the gh: prefix, I assume that the .git suffix check means that Amulet (and thus deployer) support git repo http URLs, right? So, I was thinking amulet could just convert gh: prefix URLs to https: URLs to GitHub. I do also think deployer should add the gh: prefix as well, though === kadams54 is now known as kadams54-away [23:02] cory_fu, yeah I think both amulet and deployer would need to support it, but it should be an easy add [23:02] cory_fu, of course, removing the duplication would mean that would just be a fix in deployer [23:02] cory_fu, add the support for sniffing the .git control dir [23:03] whit: I still don't understand, though. If Amulet pre-processes gh:user/foo into https://github.com/user/foo.git, why would deployer need to support it? [23:04] cory_fu, because it's not really useful for to create bundle files solely for testing [23:04] Agreed [23:04] cory_fu, ie... you could use gh aliases for just simple test stuff, but your bundle files would still have to use the other [23:04] True [23:05] cory_fu, so there is value supporting it in amulet, but it's less socially wierd if they both support it [23:05] I thought you meant it wouldn't work w/o deployer support. I agree that it's not useful w/o deployer support, though [23:05] cory_fu, but really amulet shouldn't be worrying about concerns deployer or jujuclient should handle [23:06] Indeed [23:06] but these technologies are organically grown based on need, so some pruning is to be expected [23:11] > "Less socially weird" [23:11] i like this analysis of our tooling [23:15] * marcoceppi welcomes all contributions that make amulet more or less socially weird