[00:14] on my 16.04 I tried to follow this https://jujucharms.com/get-started and failed miserably [00:14] $ juju quickstart mediawiki-single [00:14] juju: command not found [00:14] That's the step 2 [00:14] I already went through step 1 [00:14] any suggestion how to get it working? [00:44] cory_fu: looks good [01:57] axisys i think thats a typo, it should read juju-quickstart mediawiki-single [01:57] however the fact its saying juju isn't found isn't promising... [01:59] axisys if you wander back and still need help i'll be around a couple hours longer. feel free to ping and i'll help get you sorted [02:01] I'm having trouble building juju(cb347bb). Following the instructions in the README.md results in: cannot find package "github.com/Azure/azure-sdk-for-go/Godeps/_workspace/src/github.com/Azure/go-autorest/autorest/mocks" [02:03] a123 check in #juju-dev, the core hackers that are around should be able to lend a hand [02:04] thanks lazyPower === redir_ is now known as 16WAADNFF === 16WAADNFF is now known as redir === scuttlemonkey is now known as scuttle|afk [07:45] To begin juju,should I use ubuntu-server or ubuntu-desktop? [08:12] when I "juju quickstart mediawiki-single",juju gui is not in my browser [08:12] sam_yan: what version of juju are you running for a start? [08:13] 1.25.5-wily [08:13] need in ubuntu-server ? I use ubuntu-desktop [08:13] you read my DCOS tweet you aiming to get to that point? [08:14] or just general hacking around with Juju? [08:15] have not. I am a newer.I read the "install juju and deploy your services in minutes" [08:15] https://twitter.com/sam_yankun you're not that person who looks like an owl? :P{ [08:15] I am [08:16] so https://twitter.com/sam_yankun/status/724927897123852288 was you ;) [08:16] anyway! On your main computer you can run either run Desktop or Server [08:16] doesn't matter [08:17] I run my main juju controller in the data centre so I run ubuntu server but you can use desktop locally as well [08:17] it doesn't matter [08:17] also, if you're starting from scratch, grab a copy of Xenial if you can [08:17] ok . https://jujucharms.com/get-started I am learning this. [08:17] it has Juju 2.0 beta which you're better off getting started with [08:17] so my controller runs Ubuntu Server Xenial [08:18] but after "juju quickstart mediawiki-single",there is no response [08:18] okay well that should work [08:18] run juju status [08:18] does it say your stuff is up and running? [08:19] if you want a gui, run "juju deploy juju-gui" [08:19] it says "service unknow" [08:21] https://jujucharms.com/get-started it says "running this command will open the juju gui in your browser" [08:22] juju status says service unknown? [08:23] that said, it will only open juju gui once it's bootstrapped a new environment, so you need to go through the stuff you see in the terminal in that page first [08:23] i will try [08:23] sam_yan: basically, juju should ask you for AWS credentials or something, so it has somewhere to host your units [08:24] then when you fill them in, it should bootstrap a node in the cloud to run the controller, and if it or you deploy juju gui it will bootstrap another node to run juju gui [08:24] I use vmware,it matters? [08:24] dunno [08:25] I'm not sure how that quickstart stuff works I've not used it. There is also juju local which will deploy stuff to your computer as opposed to cloud services [08:25] if you want to test lcoally [08:25] where? [08:25] where what? [08:26] juju local [08:27] https://jujucharms.com/docs/1.25/config-local [08:29] it is for 12.04LTS,I use 15.10 [08:30] the docs should be the same I believe [08:30] or same-enough [08:30] ok thank you [08:37] :juju add-unit dcos-agent -n 3 it says "ERROR service "dcos-agent" not found" [08:38] sam_yan: dcos hasn't been deployed to the charm store [08:38] so it wont deploy [08:38] how can I try? [08:39] you need to checkout my 3 github repositories and setup your environment to build charms [08:39] export JUJU_REPOSITORY=$HOME/charms [08:39] export LAYER_PATH=$JUJU_REPOSITORY/layers [08:39] export INTERFACE_PATH=$JUJU_REPOSITORY/interfaces [08:39] that should pretty much do you [08:39] then inside dcos-master and dcos-agent [08:40] run charm build --series wily [08:40] juju deploy --repository=/home/youruser/charms local:wily/dcos-master [08:40] would then deploy the master for you [08:40] and [08:40] juju deploy --repository=/home/youruser/charms local:wily/dcos-agent [08:40] would deploy the agent [08:42] ok .I think the first thing to do is to be familiar with the juju,and to do as you say. thank you magicaltrout I will do as you tell me [08:42] no problem. Its not hard to build, but it helps to have a working environment first ;) [08:43] year [09:33] guys need some help. Lets say the latest availble charm revision is cs:trusty/charm-name-10 .. if i push a revision on top of it .. that revision should be accessible on cs:trusty/charm-name-11 [09:33] right? [09:33] not happening for a charm of mine [09:34] bbaqar: you using the launchpad ingestion stuff? [09:34] ie: commit to LP, wait an hour, see if its updated? [09:35] magicaltrout: that is exacly what i did .. but its been two days now .. [09:35] yeah its broken, and it looks pretty terminal [09:36] (speaking from a non canonical emplyee pov :) ) [09:36] there were a couple of posts to the mailing list [09:36] its recommended you use the new charm publish stuff to get charms into the charm store [09:36] also its instant, you don't have to wait [09:36] https://insights.ubuntu.com/2016/04/16/charm-charm-tools-2-0/ [09:38] magicaltrout: okay i ll try that out .. but can i push to any space? like lp:~/charms/trusty//trunk [09:39] it will login to the charmstore using your LP creds [09:39] so I think you end up doing the same thing [09:39] i'm not sure though, if you swing back in an hour or 3 marcoceppi will be able to offer much greater insight than me [09:45] magicaltrout :thanks alot [09:45] no probs [11:56] can anyone tell me how to get a username/password from the rabbitmq-server charm so i can connect with an amqp:// url? [12:11] I use "juju help local-provider use on this computer" and get "public-address: 10.0.3.205". But in firfox,it tells "The proxy server is refusing connections" Why? [13:31] my percona-cluster on xenial fails at creating new databases: INFO shared-db-relation-changed _mysql_exceptions.OperationalError: (1045, "Access denied for user 'root'@'localhost' (using password: YES)") [13:31] any ideas? [13:51] Hi. I had some issues with date on the server, time went forward, and since then I have logs of errors from 2032 saying "watcher has been stopped". I'm using OpenStack Autopilot and wondered about restarting watcher? [13:51] I'm not getting any logs at the moment [13:52] tail -f /var/log/juju/*.log feeds nothing back, used to be quite crazy [14:32] bdx, howdy. jewel 10.2 release dates and xenial release dates were out of alignment, but 10.2 is on its way to Yakkety, then Xenial as part of the FFe/SRU @ bug 1563714 [14:32] Bug #1563714: [FFe][SRU] Ceph Jewel stable release [14:36] bootstrapped juju environment stopped responding, juju debug-log hangs [14:50] hi, i was trying to install juju2.0 , if i do "juju status" it is getting hung [15:01] hi tvansteenburgh - aiui, rmq creates users @ amqp-relation-changed --> configure_amqp() --> create_user() ie. https://github.com/openstack/charm-rabbitmq-server/blob/master/hooks/rabbitmq_server_relations.py#L152 === redir_ is now known as redir [15:02] beisner: thanks. it turns out the guest/guest user worked, i was just trying to connect to a vhost that didn't exist :/ [15:03] tvansteenburgh, yah but that won't work forever, as it's an undesirable security gap, locking down when this gets resurrected: https://code.launchpad.net/~pjdc/charms/trusty/rabbitmq-server/no-more-guest-user/+merge/276831 [15:04] depending on the rmq version, the guest user either is or is not accessible from remote hosts, so it's kind of bad vs. bad [15:05] beisner: yah. for my use case right now it's ok [15:06] tvansteenburgh, woot. just wanted to make sure you didn't get used to that unintended feature [15:09] so I end up installting juju-2.0 and now I have juju [15:09] lazyPower: ^ [15:09] does it work with vmware vcenter or vmware vsphere? [15:10] I like to test install a hadoop cluster and we have over 20 sites I can access through vcenter.. so testing with vagrant and it works .. wanted to test with juju [15:11] https://jujucharms.com/docs/devel/config-vmware [15:12] however that is for juju-1 [15:12] hmmm, shouldn't vmware be there if I `juju update-clouds` and then `juju list-clouds`? [15:21] mbruzek https://github.com/juju-solutions/layer-tls/pull/26 [15:23] jcastro: ah ok.. I see it now [15:23] jcastro: ~$ juju-1 generate-config --show | grep -A1 vmware [15:23] vmware: [15:23] type: vsphere [15:23] yeah it's just in 2.0 it should set that up automatically, you shouldn't be mangling the config file by hand [15:24] I know it works in juju-1 [15:24] jcastro - i dont see vmware mentioned in the release notes for beta-5 [15:24] nor me [15:25] picky me.. would be nice if there were a bash completion of the juju arguments .. like git has [15:25] it does have that normally [15:25] axisys: there will be, for juju 2.0, when it's released iirc [15:25] I don't know why it's currently broken, I just noticed that today [15:28] bcsaller cory_fu I want to use a custom tactic in my layer. how can I? [15:29] why would this charm https://jujucharms.com/u/charmers/block-storage-broker/trusty/2 be at lp:~charmers/trusty/block-storage-broker/trunk but not at lp:charms/trusty/block-storage-broker ? [15:29] nottrobin: because it's not promulgated [15:29] hmm.. juju quickstart fails, what am I doing wrong? http://dpaste.com/20N9MQN [15:29] Ostensibly, you put the tactic in your layer, and add it to the "tactics" key in layer.yaml. However, lazyPower ran into an issue with how `charm build` loads the tactics: https://github.com/juju/charm-tools/pull/191 [15:29] marcoceppi: ^ [15:30] axisys: what does `juju version` say? [15:30] I picked automatically create and bootstrap a local enviroment [15:30] 2.0-beta4-xenial-amd64 [15:30] marcoceppi: ^ [15:30] axisys: you don' tneed quickstart for 2.0 [15:30] what in gods name is a tactic?! [15:30] * magicaltrout cracks out the juju dictionary [15:31] marcoceppi: Here's the example using the tactic in a layer: https://github.com/juju-solutions/layer-docker/pull/43/files Note the docstring [15:31] axisys: just bootstrap an enivronment, then juju deploy mediawiki-single [15:31] marcoceppi: I need a get started for 2.0 :-) [15:31] marcoceppi: ok [15:31] axw: looks like you touched the vmware provider last in march, do you know why it wouldn't be showing as an option in add-clouds or list-clouds? [15:31] axisys: yes https://jujucharms.com/docs/devel/getting-started [15:32] axisys: 2.0 is still technically in development [15:32] magicaltrout: tactics are used in charm build to determine how to handle files/merging [15:32] jcastro: I decided to come back.. dont have same environmen anymore.. [15:32] * jcastro nods [15:34] cory_fu: my django layer had a django.yaml, which I'm converting to layer.yaml, so I'd like to basically have a tactic that, if django.yaml is found it just yells at the user to migrate [15:36] marcoceppi: Testing that PR on charm-tools would be much appreciated. But the main thing you need to worry about without it is that your imports at the top of the file containing the tactic likely won't be visible when your tactic is actually invoked. lazyPower could tell you if there are other issues, and I know he ended up removing the custom tactic in the end, so there might be [15:36] cory_fu - only until it gets merged into charm-tools [15:36] Hrm [15:37] cory_fu - that patch that you linked worked. and i commented on teh commit itself as thats what i had to go on [15:37] * lazyPower will comment on the bug [15:37] lazyPower cory_fu thanks for the feedback, I left the merge because I didn't see much feedback/way to test [15:38] also, no one checked of the checklist [15:38] sorry [15:38] preoccupied with a mountain of things to try to land by EOW [15:40] lazyPower: no worrie [15:46] marcoceppi: Did you set that checklist up as a template for PRs or something? [15:46] cory_fu: yes [15:46] Where do you set that up? That's pretty awesome [15:47] https://github.com/juju/charm-tools/tree/master/.github [15:47] Oh, nice [15:47] cory_fu: I wanted to experiment with the charm-tools repo before pushing it out elsewhere [15:47] esp since both charmstore-client and charm-tools are confusing [16:47] is baking the gui into the bootstrap node still planned? [16:47] or did that get abandoned [16:50] LiftedKilt: already done I guess. [16:50] "juju gui" works out-of-the-box with 2.0beta-something [16:53] ReSam: oh snap cool [16:53] I must have missed that [16:55] creating models is still broken from the gui though haha [16:58] LiftedKilt: what version of juju? [16:58] LiftedKilt: are you using the baked in gui - because that totaly works [16:58] beta6 [16:58] using baked in gui [16:59] open drop down, type in name of model, click new -> it just closes the drop down but doesn't create anything. It has never worked for me though [16:59] in firefox or chrome === rcj` is now known as rcj === rcj is now known as Guest12905 === Guest12905 is now known as rcj [17:21] any of y'all written a charm with resources yet? [17:22] looking for an example to learn from [17:22] cmars - its not complete, but i've been looking at adding it to layer-tls so easyrsa is not beign pulled in via git [17:22] lazyPower, what's a good way to locate the resource file after resource-get from within the charm? [17:23] that i dont know :) i haven't gotten further than declaring it exists [17:23] lazyPower, maybe os.path.join(hookenv.charm_dir(), '..', 'resources', resource-name) ? [17:23] ok, i'll play with it [17:23] that sounds like a charm-helper waiting to be written [17:23] and share if i get it working [17:23] yeah [17:23] or a charms.module [17:23] cmars - much <3 if you do. I'm eager to see how you solve it [18:11] is there a way to just watch the output of [Units] from juju status? [18:11] i've been doing: [18:11] watch "juju status | sed -n '/\[Units\]/,/\[Machines\]/ p' | head -n -2" [18:11] but wondering if there's a more elegant way [18:12] Hi [18:13] I am facing some issue while running amulet test, Can anyone please help me! [18:13] 2016-04-27 17:59:45 DEBUG juju.worker.logger logger.go:45 reconfiguring logging from "=DEBUG" to "=WARNING;unit=DEBUG" 2016-04-27 17:59:46 INFO install /usr/bin/env: python3: No such file or directory 2016-04-27 17:59:46 ERROR juju.worker.uniter.operation runhook.go:107 hook "install" failed: exit status 127 [18:13] exit [18:13] LiftedKilt: not as such, but I wonder what `juju status "*/*"` displays [18:14] mgz: [Services]: command not found [18:14] suchvenu: you don't have python3 installed but your install script uses it [18:14] yes i have it [18:15] Hi, quick question, how to debug juju charms? [18:15] suchvenu: locally, but not on the machine that's being run? ssh in and poke around [18:16] i have written in my 00-setup file [18:16] cat << EOF > local.yaml ibm-repo: db2_curl_url: "$DB2_CURL_URL" db2_curl_opts: "$DB2_CURL_OPTS" db2_licese_accepted: "$DB2_LICENSE" #db2_package_name: "$PACKAGE_NAME" EOF #is sudo required here ? sudo add-apt-repository ppa:juju/stable -y sudo apt-get update sudo apt-get install amulet python3 -y [18:17] SaltySolomon: https://jujucharms.com/docs/stable/authors-intro and so on [18:19] suchvenu: I don't see how that helps [18:19] mgz: I got one of the default packages from the charms store and keystone is not working [18:23] SaltySolomon: you can pastebin the yaml status, and the unit log from keystone (get with juju ssh keystone/0) into a mailinglist post or bug on launchpad [18:26] suchvenu: for your install hook to work, you need use a scripting language that's pre-installed on the default image [18:26] suchvenu: old ubuntu series didn't have python3 by default [18:27] suchvenu: so either your charm needs to say it's xenial only, or you need to make the install script bash that installs python 3 then invokes from there [18:27] sorry was away [18:28] ok... I see [18:28] let me check [18:31] mgz : So i need to specifically install python3 in my install hook, right ? [18:38] suchvenu: depending on the series of you charm, yes. and obviously using any kind of python to do that install hook is therefore a problem (for multi-series charms) [18:38] i am using trusty [18:38] and i am writing as a layered charm [18:39] then just change "#!/usr/bin/env python3" to "#!/usr/bin/env python" and be done [18:39] or... whatever with layered charms, bug one of the maintainers [18:43] Its getting deployed manually, some env issue with amulet [18:43] i am trying to change python3 to python as you said [18:43] might be something we need marcoceppi for then [18:45] no, it didn;t help with that [18:47] suchvenu: send a mail to the juju list, I think everyone's at ODS this week [18:47] ok [18:47] will do [19:09] mgz: what's up? [19:11] marcoceppi: suchvenu was having issues writing amulet tests that looked like something was assuming python3 is installed on trusty [19:39] cory_fu, do you have time for a quick question? It's about ignoring files in interfaces when they get copied as part of reactive charms. [19:39] Sure [19:40] So, 'we' (as in OpenStack charmers) are trying to add tests to interfaces, but they get copied into the final charm and then blow up on install (due to missing python modules). ... [19:41] cory_fu, so I looked at the code and found an 'ignore' key, but it only seems to work for layers (or just the top level charm). [19:42] cory_fu, I stared at the code and saw that during the interface copy it uses the target_config of the layer that brings in the interface, but doesn't use any config from the interface. [19:42] tinwood: Did you try adding the "ignore" key to interface.yaml? [19:42] cory_fu, so I was wondering how hard it would be to change it so that we could ignore files from the interface during the copy? [19:42] cory_fu, yep, it's ignored. They only seem to work in layers. [19:42] Hrm [19:44] It's due to the target_config in plan_interfaces() using the layer rather than the interfaces' yaml. (I think). [19:44] cory_fu, (which is probably for a very good reason!) :) [19:47] tinwood: I don't know. It seems like it should honor ignores in the interface.yaml as well. InterfaceCopy should still have access to those ignores, via self.interface.config.ignores, so I think we might just need to add that list to the target config's list at https://github.com/juju/charm-tools/blob/master/charmtools/build/tactics.py#L159 [19:47] bcsaller: Can you chime in whether there's a reason that only the charm layer's ignores are honored? [19:47] Or if it's just an oversight [19:48] cory_fu, yes, I wasn't sure where the best place was to add it, nor whether there was an important reason to not do it; it seemed that the framework was there to make it work. [19:49] tinwood: I think we should also have an additional set of default ignores for interface layers. E.g., I don't think every interface layer should have to explicitly exclude tests [19:50] cory_fu, right. Yes, that makes sense. [19:50] mbruzek if you have a sec, i could use eyes on this one - https://github.com/juju-solutions/charms.docker/pull/18/files [19:51] cory_fu: most likely an oversight, I can't recall an intent [19:51] cory_fu, it would be great if 'tests' was added to DEFAULT_IGNORE list as that could create a template for how to add unit tests to an interface or layer. [19:51] We don't always want to exclude tests from every layer, though, do we? [19:51] cory_fu, however, that might be going to far! [19:52] Seems like it would make sense for interface layers, as they won't get picked up anyway. But base layers could potentially add amulet tests to the final charm, or something [19:52] cory_fu, no you're probably right. I can see the source charm for the build bundling tests for the final built version that can be exercised by the CI as a gate. So yes, it would be foolish to ignore them by default. [19:55] cory_fu, bcsaller, how would you like to proceed? Shall I raise a bug? I can also (try to) to do a pull request if that helps? [19:56] tinwood: Yeah, that would be good. I'm optitmistic that it'll be a simple change at the line I linked above [19:57] thats what they all say.... [19:57] ..... 6 days of hacking later [19:57] cory_fu, okay, I'll try it and let you know. :) [19:57] lol, too true, magicaltrout [19:58] magicaltrout, indeed. Down the rabbit hole one proceeds ... [20:02] marcoceppi: juju.fail’s server DNS address could not be found. :( [20:03] natefinch: oops, that'll be addressed in a hot min [20:04] marcoceppi: forget to pay your registrar again? ;) [20:04] natefinch: honestly, so many domains across so many registrars, a few have old credit cards [20:04] marcoceppi: I hear ya [20:04] its like a startup all over again! [20:05] marcoceppi: I'm lucky I only have domains on two registrars... I'd love to migrate off godaddy, but it's such a hassle. That's how they get you, I guess. [20:06] natefinch: melbourne it, tucows, godaddy, google domains, iwantmyname, ghandi [20:06] natefinch: those are the ones I remember [20:06] marcoceppi: ouch [20:07] marcoceppi: million dollar idea - service that migrates your domains for you. Actually... I'm surprised the bigger registrars don't code something up to make it easier. [20:07] natefinch: it's pretty straight forward, but it still requires human intervention [20:07] natefinch: and yo0u have th pay for another year on xfer [20:07] that'd be like me dolling out a few grand to consolidate [20:08] a few grand to consolidate sounds like a bargain [20:08] * magicaltrout goes to setup such a service [20:09] if letsencrypt can figure out a way to make getting a friggin' cert automatic, we should be able to figure out how to transfer a stupid domain [20:12] natefinch: juju.fail should be back unless you've got dns cache [20:16] marcoceppi: it seems I do [20:33] natefinch: if it's still not owrking, try a different browser [20:34] or reset your router :) [20:34] cory_fu, bcsaller: the good news is that your suggested change of 'self.config.ignores' -> 'self.interface.config.ignores' works to pick up the ignore config option in interfaces.yaml. I'll do a PR. [20:36] tinwood: Make sure you also include the existing ignores. It should be `self.config.ignores + self.interface.config.ignores` [20:36] woooo \o/ [20:36] Glad to hear it was easy, tho [20:37] cory_fu, they'll be dups then as both lists contain the defaults (I'm guessing). Shall I list(set()) them to get a unique list? [20:38] tinwood: self.config.ignores would contain the "ignores" key from the top (charm) layer's layer.yaml, while self.interfaces.config would contain the "ignores" key from interface.yaml [20:39] cory_fu, I printed them both and they both contain the DEFAULT_LIST: ... [20:39] The ignore list is: ['.bzr/', '.git/', '**/.ropeproject/', '*.pyc', '*~', '.tox/', 'build/'] [20:39] The self ignores list is: ['.bzr/', '.git/', '**/.ropeproject/', '*.pyc', '*~', '.tox/', 'build/'] [20:39] * tinwood duh, wrong one: [20:39] charmtools.build.tactics: Copying Interface keystone: /home/ubuntu/charms/trusty/designate/build/builds/designate/hooks/relations/keystone [20:40] The ignore list is: ['.bzr/', '.git/', '**/.ropeproject/', '*.pyc', '*~', '.tox/', 'build/'] [20:40] The self ignores list is: ['tests', '.bzr/', '.git/', '**/.ropeproject/', '*.pyc', '*~', '.tox/', 'build/'] [20:45] tinwood: It shouldn't matter if it has dupes [20:46] cory_fu, okay, I'll concat the two vars. Thanks very much for your help! Typically, how long will this take to land (assuming it's accepted)? [20:47] marcoceppi: ^ [20:50] cory_fu: IM CONFUSED [20:50] im also caps locked [20:59] is there a way to skip installing g++ in the basic layer? a configuration option, perhaps? [20:59] marcoceppi: If a PR is accepted on charm-tools, how long until it will be available (somehow, e.g., ppa)? Or are we still too new to the release process? [20:59] most of my charms don't need it and some ops folks are hesitant to deploy compilers into a production env [21:00] cory_fu: I'm working on getting an MRE, which means we can turn around a micro release (patch) in a copule of days [21:00] as for me, i think it just slows down the install hook [21:00] cmars: it's typically installed for python-dev === natefinch is now known as natefinch-afk [21:00] marcoceppi, can i not install python-dev? [21:01] cmars: maybe? but not atm [21:01] ok [21:01] cory_fu: ^? [21:34] cory_fu, marcoceppi thanks for the info. we'll work around the problem with a post build script until the patch (I've not yet submitted!) lands. [21:34] gnuoy, ^^^ - found issue, have solution, but need to keep the delete workaround until it lands (... this is for tests appearing in the built charm). [21:41] aisrael: I'm seeing some strange behavior when trying to use the watch-files options of rsyslog-forwarder-ha in a bundle. If I don't tag the revision explicitly, I get an error saying the option is invalid, but if I explicitly give it -8, it works. Any idea why? [21:41] aisrael: Did you by chance use `charm publish` for rev 8? [22:02] jcastro: IIRC you have to set a feature flag for vsphere to show up [22:04] axw: yeah alexis had me file a bug on that, thanks [22:05] np [23:03] cory_fu: I don't think that's one I used charm publish for. The first one I used that on was auditd [23:05] Very strange