=== jcsackett_ is now known as jcsackett === redir_sprint is now known as redir === ahasenac` is now known as ahasenack === ahasenack is now known as Guest64506 [07:54] good morning juju world! === frankban|afk is now known as frankban === spammy is now known as Guest95230 [08:52] Hi everyone, a member/partner from scaledb cannot register with the juju list. Who is the juju list owner? [10:17] Hello there, i'm looking for the minimum requirements for a juju "node"/vm but i can't seem to find it [10:23] BlackDex: What do you mean by minimum requirements? [10:24] BlackDex: A base charm is the ubuntu charms [10:24] BlackDex: That deploys ubuntu and it should run wherever ubuntu runs [10:29] I meen the juju bootstrap node [10:29] So, how much mem, disk, cpu etc.. is the minimum requirement for a juju-bootstrap node [10:30] For maas it is located here: http://docs.ubuntu.com/maas/2.1/en/#minimum-requirements [10:30] For JuJu i couldn't find it [11:45] Is there any equivalent of optional_resources from jujuresources in Juju 2.0 resources? [11:45] I'm trying to write a charm that installs packages from resources if they're provided that way, but otherwise installs from a PPA [11:46] Simply not attaching the resources and having a bit of conditional behaviour in the charm worked OK until I tried to do "charm release", at which point I got "cannot release charm or bundle: bad request: charm published with incorrect resources: resources are missing from publish request: ..." [11:52] I've thought about building metadata.yaml conditionally, but that's fiddly, and it would mean two different versions of the charm, one with resources and one without [11:53] Do I just have to fall back to bundling files in the charm instead? Seems a shame - resources are a nicer UX in general [12:04] Optional resources.... [12:05] you can have dummy 0-sized recources [12:05] True; feels extremely hacky [12:05] cjwatson: there is an email thread that is somewhat related [12:05] Is that something that other charms do? [12:05] perhapse you could give some feedback [12:06] yes, there is no other option for now [12:06] but your feedback/usecase needs to be heard [12:06] please send some email to the list/or say something about your needs in that email thread [12:07] cjwatson: title "Hide resources from showing on the store" [12:07] * cjwatson finds https://lists.ubuntu.com/archives/juju/2016-December/008265.html [12:08] kjackal: do you happen to have the Message-Id of that message? That way I can synthesise a reply - I'm not subscribed to that list [12:14] kjackal: ah, never mind that, I can extract it from the HTML source [12:29] sorry cjwatson. I am back now === freyes__ is now known as freyes [12:52] Hi all, iam trying to push multiseries charm to charm store. i have successfully ran charm push but charm publish command if giving following error http://paste.ubuntu.com/23583217/, anyone please suggest? [13:17] Hi all, iam trying to push multiseries charm to charm store. i have successfully ran charm push but charm publish command is giving following error http://paste.ubuntu.com/23583217/, anyone please suggest? [13:18] hi shruthima I thought the publish was replaced by charm release [13:19] i get a "ERROR unrecognized command: charm publish" [13:19] shruthima: is it possible you are running some old version of charm tools? [13:19] ya in some machines i faced even that error [13:20] iam using charm-tools 2.1.2 [13:21] kjackal: any document related to charm release? [13:22] should be the same as charm publish [13:22] oh k il give a try [13:23] kjackal: ERROR unrecognized command: charm release [13:24] in which version of charm-tools will it work? [13:24] I couldnt find docs on the web for charm release, but here is the --help output: http://pastebin.ubuntu.com/23583338/ [13:25] shruthima: Regarding the "access denied" error you are getting I would start by looking at the permissions you have on launchpad [13:26] shruthima: are you in the right group(s)? [13:26] kjackal: ya i have pushed 3-4 charms with same launchpad id [13:27] so it is only that one you cannot release? [13:27] you can release other charms in that group? [13:28] almost everyone is facing this issue from last friday [13:28] In that case, use charm show to compare the permissions set [13:28] last friday... [13:28] strange.. [13:29] let me try to release a charm myslef. [13:29] kjackal: no other charms also iam unable to publish now [13:29] ok [13:33] shruthima: I am sorry I cannot reproduce the error. Releasing charms works here: http://pastebin.ubuntu.com/23583368/ [13:34] shruthima: can you try the publish command with the --debug option [13:34] kjackal: almost everyone from ibmcharmers facing same error http://paste.ubuntu.com/23583370/ [13:34] ok [13:35] shruthima: you have an old version of the charm command, if you run `charm version` you should get charm with at least 2.2.0 [13:38] marcoceppi: how to update the charm version [13:38] `sudo add-apt-repository ppa:juju/stable; sudo apt update; sudo apt dist-upgrade` [13:39] `sudo add-apt-repository -y ppa:juju/stable; sudo apt update; sudo apt dist-upgrade` [13:39] kjackal: here is the output with debug option http://paste.ubuntu.com/23583387/ [13:41] shruthima: can you please try the upgrade marcoceppi suggested and then do a charm logout, login, release ? [13:41] kjackal: ya iam trying to upgrade now [13:41] shruthima: thanks [13:52] marcoceppi: kjackal : i have updated the cahrm version http://pastebin.ubuntu.com/23583438/ , still iam facing the same ERROR cannot release charm or bundle: unauthorized: access denied for user "salmavar" [13:52] shruthima: which charm are you trying to release? [13:52] ibm-im [13:53] multiseries charm [13:53] shruthima: you can't release to the stable channel for that charm, it's an approved charm. [13:53] https://jujucharms.com/ibm-im/ [13:54] instead, sonce you push, enter the URL you get back from push into http://review.jujucharms.com/ and it'll follow a review cycle then be released once reviewed [13:56] marcoceppi: oh ok so just i have to create a new request for review with new URL right ? [13:56] yes [13:56] marcoceppi: thanku :) [13:58] marcoceppi: any updates on this issue https://github.com/juju-solutions/interface-http/issues/5 [14:16] shruthima: no, not at the moment [14:18] marcoceppi: ok when can we expect that to be resolved? [14:22] is there a way, when adding a machine manually with 2.x, to define what kind of lxc networking you'd like to use by default? === MmikeM is now known as Mmike [14:30] marcoceppi: Hi, i joined recently the aws charmers program, I was wondering if i do have access to AWS web console as well. [14:30] deanman: you do not [14:31] deanman: is there something you need in the web console? [14:32] marcoceppi: i was controlling a deployment yesterday from home and now from office (and behind proxy) it seems my client cannot talk to AWS. I'm trying to debug whether is due to controller dies or some other network issue. [14:32] deanman: ah, do you know the instance id for your controller? [14:33] marcoceppi: does that help ? http://paste.ubuntu.com/23583601/ [14:34] deanman: ah, can you get the controller UUID from `juju show-controller` [14:36] marcoceppi: hmm it seems that i can't http://paste.ubuntu.com/23583609/ [14:37] deanman: $HOME/.local/share/juju/controllers.yaml should have that cached [14:39] marcoceppi: http://paste.ubuntu.com/23583631/ [14:39] uuid: d1a79bb1-a362-4551-81ec-d51613d31a1f [14:40] deanman: it is alive still [14:40] bootstrapped about 22 hours ago [14:41] marcoceppi: how does local juju client talks to the controller ? ssh or other protocol ? SSH is blocked and needs some tsocks configuration but everything else should go through the proxy server. [14:43] deanman: I'm under the impression that it's directly with a websocket [14:43] ah websockets [14:44] ok thanks, ill try troubleshoot it === rye is now known as ryebot [15:41] goood morning [15:41] are there any docs at running a juju controller in HA? [15:41] im designing an HA MAAS solution and haven't been able to find any docs yet [15:41] bildz: let me take a look [15:42] marcoceppi: thanks! [15:45] bildz: in short, `juju enable-ha` is the command you want (juju help enable-ha for more info). I've filed a bug on the docs to document this better https://github.com/juju/docs/issues/1556 [15:47] Hi Kevin [15:48] Hi i have firefox browser issue on ubuntu z machine ,can anyone help me [15:49] Ankammarao: you mean, a problem with Firefox on Z or problem with Juju GUI on Firefox (on Z)? [15:50] when running firefox command on vnc server ,getting the error like "Error: GDK_BACKEND does not match available displays" [15:50] marcoceppi ,i want to open firefox browser on vnc session for ubuntu z machine [15:51] Ankammarao: this sounds like a general Ubuntu question. http://askubuntu.com or #ubuntu on irc [15:51] Ankammarao: those are probably the best source for help [15:51] Ankammarao: I'd help, but I've not encountered that problem :( [15:55] marcoceppi, can we install other browsers like "firefox" on ubuntu z machine [15:55] Ankammarao: probably, sure. It's the same ubuntu as you'd expect on any other architecture. So long as there's a package for that architecture [15:58] bildz: https://jujucharms.com/docs/2.0/controllers-ha [16:08] petevg: hey, are you in DC? [16:12] marcoceppi that I am. [16:12] Just for the day. [16:32] petevg: you visiting the whitehouse before it turns gold? [16:32] personally, I think it'll look nice gold... [16:35] magicaltrout: heh. I thought about doing that. [16:35] I have just enough time to wander through the National Mall, though. Then I have to hit the conference. [16:35] i was down there the other day.... huge blisters [16:35] I never learn [16:36] Outside > inside. I'll have to see the Whitehouse after Trump's successor restore it :-) [16:36] I reckon he'll turn it into a downtown golf course [16:36] he'd make a lot of money..... [16:36] Ew. [16:36] think about it [16:36] teeing off from the roof, with the hole being the washington monument [16:36] it doesn't get any better than that [16:37] Sigh. [16:37] you could zipwire from the roof [16:37] The problem is that I can see it. [16:37] then, tee #2 could be the lincoln memorial [16:37] it would be the best golf course ever [16:38] i need to figure out how to incorporate capitol hill [16:41] I also happen to think the roof of the kennedy centre would make a great green [16:41] plenty of room, but testing enough because if you go off the edge, you're gonna have a tricky chip to get back onto the green from 4 stories below [16:41] maybe a short par 3 from the Watergate hotel [17:09] marcoceppi: can you link me to anything you have done for datadog-agent? [17:18] marcoceppi: https://github.com/marcoceppi/svg.juju.solutions/pull/8 [17:48] bdx: give me 10 mins to clean up this iteration [17:48] bdx: and we can chat [17:48] marcoceppi: ok [17:49] cory_fu: thanks for the contribution! [17:49] cory_fu: I wnat to merge post/get, since it's a lot of duplication [17:49] maybe one day [17:52] rick_h: you were talking about how you specified the default aws region in clouds.yaml, how exactly are you specifing that? [17:53] bdx: don't listen to rick, use juju set-default-region ;) [17:53] cory_fu: I've got a crazy idea, need sanity check [17:54] cory_fu: got a min? [17:54] marcoceppi: Sure [17:54] marcoceppi: thx [17:55] cory_fu: https://appear.in/lets-talk-about-interfaces-charms-and-reactive [17:59] bdx: yea right , that Rick guy is nuts. Don't listen to him. [18:09] hey guys, I'm from NGINX and working on the official charm, does someone have time to take a look at it and see if I'm going in the right direction? [18:15] lucacome: yeah, totally [18:19] marcoceppi, I've started form layer-nginx and continued from that https://github.com/lucacome/juju-layer-nginx [18:21] lucacome: taking a look now! [18:21] thank you [18:21] lucacome: cert/key might be better specified as config options [18:23] bdx, can I pass files wit config options? [18:26] lucacome: I don't think so, you can pass the cert/key as strings <- this might not be the prefered way either tho [18:26] luacome: https://github.com/jamesbeedy/charm-datadog-agent/blob/master/config.yaml#L22 [18:28] lucacome: others have made charms and layers to help facilitate SSL/TLS, see https://jujucharms.com/u/containers/easyrsa/3, and https://github.com/cmars/layer-lets-encrypt === frankban is now known as frankban|afk [18:29] lucacome: ^ only cover self signed, and LE certs/keys though ... so to acommodate user specified sighned cert/key, I would thing config options would do it, but I'll let the pros chime in here [18:30] signed* [18:30] think* [18:31] bdx, they are not self signed cert/keys, they are handed out by us to install the NGINX Plus, our commercial version [18:32] lucacome, I see that now [18:32] but I'll take a look in to it to see how others are doing dealing with certs [18:33] cory_fu: I've backed myself into a corner [18:33] lucacome: I would lean towards config options for that .... @marcoceppi should be able to confirm whether that is the best route or not [18:34] lucacome bdx I don't think resources is a bad way, but if you had a "nginx-cert" resource that was a tar.gz of a .key and .crt file it'd be easier for users [18:34] I'd go either way, config or resources - both are valid [18:35] marcoecppi: in his charm the resouce is optional though [18:35] I do have a few other concerns, still working through it [18:35] bdx: yeah, and that's fine, as long as there's a 0 byte blank resource [18:35] ahhh [18:35] neither config or resources is ideal for this [18:35] cory_fu: can I do something like this? [18:36] bdx: haven't forgotten about you, but I have some code I think you'll like [18:36] bdx: for datadog [18:36] marcoceppi, is there a better way to upload files? [18:36] nicccee [18:37] lucacome: configuration, resources, and scp are the three methods. With the last one not being recommended at all given the UX around it [18:37] resources is preferable, because it doesn't expose the key and crt files in plain text anywhere [18:37] unless you want the configuration options for nginxplus to be exposed [18:38] in exports [18:38] * bdx eats his words [18:38] well, configuration isn't bad bdx, there are cases you'd wnat the crt to be in a bundle export [18:39] but I'm not sure that's the case here [18:39] but I also can't tell, those are the two paths ultimately [18:40] cory_fu: http://paste.ubuntu.com/23584586/ line 13, can I use getattr on the class for autoaccessors? [18:40] cory_fu: or is there an easier way? [18:48] marcoceppi: I think that's the best way, though "rel_data.k()" won't work [18:48] cory_fu: right, so will getattr work? [18:49] or is there a better way? [18:49] marcoceppi: Actually, what about standardizing the interfaces so that they share a common method name. Something like, "get_configuration()"? [18:50] cory_fu: not a bad idea [18:50] cory_fu: I want to ultimately standardize these stats interfaces === niemeyer_ is now known as niemeyer === FourDollars_ is now known as FourDollars === mpontillo_ is now known as mpontillo === hazmat_ is now known as hazmat === shawniverson is now known as spammy [20:01] is there a debug method of sorts that will list all flags/states that are set for a charm? [20:30] how do I get tests to run against a live deployment? I can't only use containers because I'm using a snap and snaps have a problem installing in containers at the moment [20:31] so my one test at the moment makes a deployment and checks that my charm is a blocking state while it waits for a snap resource [20:31] I can verify that [20:37] bdx: `charms.reactive get_states` I think [20:37] bdx: I am finally ready to talk about datadog === Mmike is now known as Mmike_1 [21:08] marcoceppi: all ears [21:08] bdx: so, this is what I came up with [21:09] https://github.com/silph-io/layer-datadog/blob/master/reactive/datadog.py#L52 [21:09] I'm testing it now, but basically it means you only need to touch metadata.yaml and layer.yaml to add new integrations [21:10] that was the biggest concern when talking to ilan at data dog, about overhead of maintenance and boilerplating [21:11] bdx: I'd like to collaborate going forward, and get this repo into upstream [21:11] marcoceppi: I'm trying to get something official going for it before I start putting it out everywhere for sure [21:11] bdx: yeah, we've been in touch with datadog about this, we have an internal business relationshiop being developed [21:12] marcoceppi: so why list integrations in metadata? [21:14] bdx: well, interfaces should drive the entire configuration for everyhting. So if the nginx and php-fpm (and kubernetes, mysql, etc) charms and layers include a unique stat inferface, which supplies the keys required by datadog, or any other stat generations/configuration tool, you get a fully automated experience [21:14] we can jump on a quick call to explain more if you'd like [21:14] marcoceppi: ok [21:15] bdx: https://appear.in/datadog [21:57] bdx: I found a few foul ups, updating the repo in a min with some patches [21:57] lots of missing imports [22:19] I have the latest juju and MAAS. When trying to "conjure-up openstack" and running the OpenStack option, I use the "Connect to an existing MAAS" option. For maas server, I am entering "http://x.x.x.x/MAAS/api/2.0/" and entering my maas-auth key in that field. When I Confirm the options, it shows that it is trying to bootstrap the environment and then almost immediately failes with "error: flag provided by not define: --upload-tool [22:19] s" does anyone know what is causing this issue? [22:24] xplatform12: to start, you should need to enter "http://x.x.x.x/MAAS/" [22:25] xplatform12: what version of juju do you have installed? [22:25] xplatform12: `juju version` [22:25] cory_fu: I hit a snag [22:26] marcoceppi: ? [22:27] cory_fu: `RelationBase.from_state('php-fpm')` returns None despite these states: {'apt.installed.apt-transport-https': None, 'datadog.configured': None, 'apt.installed.datadog-agent': None, 'php-fpm.available': {'conversations': ['reactive.conversations.php-fpm:3.silph-web/0'], 'relation': 'php-fpm'}} [22:28] marcoceppi: The states have to match exactly. You'll want RelationBase.from_state('%s.available' % relation_name) [22:28] cory_fu: oh, boss, even easier [22:28] this removes my entire `relation_name.available` state check [22:30] bleh [22:30] ? [22:31] cory_fu: http://paste.ubuntu.com/23585512/ [22:31] these are set to scope.UNIT [22:31] marcoceppi: I tried with just http://x.x.x.x/MAAS/ and my key and got the same error. I am using juju Version: 1:2.0.1-0ubuntu1~16.04.4~juju1 [22:31] xplatform12: huh, yeah, conjure-up should work just fine. stokachu ^ ? [22:32] marcoceppi: Yeah, if it's set to scope UNIT, that means there will be a separate conversation per remote unit, so you have to iterate over self.conversations() instead of assuming a single self.conversation() [22:33] cory_fu: yeah, it's a subordinate relation, so, single conversation makes sense [22:33] I also checked in /var/log/openstack.log and it has the following: "Dec 5 22:28:27 juju openstack: [ERROR] ['error: flag provided but not defined: --upload-tools']" I can't seem to find any info about htis issue online. [22:33] marcoceppi: So if it's always going to be subordinate, you can just change it to GLOBAL scope [22:33] xplatform12: --upload-tools was a flag for juju 1.x conjure up thinks that's your version, for whatever reason [22:33] cory_fu: ack, just did [22:34] time to re-deploy [22:36] marcoceppi: This was upgraded from juju 1.x. Let me rebuild the juju system and go right to 2.0 without an upgrade. I hope that helps and thanks. [22:47] marcoceppi: I rebuild and made sure I had the correct repositories when installing juju and now conjure-up is past that issue and bootstrapping. Thanks for your assistance. [22:48] xplatform12: no worries! sorry you got caught up in that transition [22:48] bdx: alright, finally got configuration files writing [22:48] marcoceppi: niceee [22:49] bdx: my latest work was just pushed to the repo [22:49] marcoceppi: looking [22:51] bdx: bunch of awful little papercuts === danilos` is now known as danilos