| stub | marcoceppi: You are infecting layer.yaml with that awful json schema stuff? Is there some gui that needs to drive this? | 00:15 |
|---|---|---|
| marcoceppi | stub: no, but build time validation is still quite valuable | 00:46 |
| marcoceppi | and this has been in charm-tools for quite a while ;) | 00:47 |
| stub | T_T | 00:51 |
| marcoceppi | stub: either way, we can't have layers hoping keys will exist, or throwing run time errors during deploy, jsonschema makes it so we can validate layer configurations at build time, before a deploy even happens | 00:52 |
| marcoceppi | stub: just trying to make sure WE HANDLE THE UNEXPECTED | 00:52 |
| stub | yeah, I just hate the spec. I mean, your schema *could* just be an example snippet that is decoded and introspected for all the type information you need. But nooo, lets repeat the mistakes of the past like XML schemas. | 00:55 |
| === natefinch-afk is now known as natefinch | ||
| === thumper is now known as thumper-afk | ||
| === lifeless_ is now known as lifeless | ||
| kjackal | Hello Juju World! | 06:34 |
| === frankban|afk is now known as frankban | ||
| === TheRealMue is now known as TheMue | ||
| magicaltrout | stub: ping | 08:14 |
| === thumper-afk is now known as thumper | ||
| stub | magicaltrout: pong | 08:42 |
| magicaltrout | ah hi stub sorry to bother you had a quick nagios question for my demo tomorrow, I was hoping you might be able to shed some light on | 08:44 |
| magicaltrout | https://github.com/buggtb/dcos-master-charm/blob/master/reactive/dcos_master.py#L200 | 08:44 |
| magicaltrout | does that look correct to you? | 08:44 |
| magicaltrout | it seems to execute fine | 08:44 |
| magicaltrout | but Nagios never picks up the service | 08:44 |
| stub | I use the nrpe-external-master relation, so not sure if that has anything to do with it. | 08:45 |
| stub | I also tend to use charmhelpers, rather than the helper on cmars' interface | 08:45 |
| magicaltrout | https://pythonhosted.org/charmhelpers/api/charmhelpers.contrib.charmsupport.html | 08:46 |
| magicaltrout | that one? | 08:46 |
| stub | yes, buried way down in contrib. That one. | 08:46 |
| magicaltrout | okay | 08:46 |
| magicaltrout | cool | 08:46 |
| magicaltrout | thanks | 08:46 |
| magicaltrout | I see all you lot using the external interface but for the demo I'd like to keep it all inside one model if possible | 08:47 |
| stub | magicaltrout: I can't see anything obvious wrong with yours. It seems wired up the way I think it is supposed to be if you are using the interface. | 08:49 |
| magicaltrout | yeah it seems like there is no rsync action | 08:49 |
| magicaltrout | never mind I'll keep prodding | 08:49 |
| stub | I'm no nagios expert. I just got sick of seeing the same boilerplate everywhere and decided to not perpetuate it. | 08:50 |
| magicaltrout | hehe | 08:50 |
| PCdude | any MAAS users/experts here, the MAAS channel itself has been dead the last 12 hours | 08:53 |
| magicaltrout | i fiddled with it for the last couple of days | 08:53 |
| magicaltrout | so I'm not an expert | 08:53 |
| magicaltrout | but i did get bits of it working | 08:53 |
| PCdude | which version did u used? | 08:54 |
| magicaltrout | 2.0-Rc2 | 08:54 |
| PCdude | yeah, the 2.0 version is good, but since the JUJU version is still in beta I decide to use the 1.9.4 version on ubuntu 14.04 | 08:54 |
| magicaltrout | I've used 2.0 since the juju alphas | 08:55 |
| magicaltrout | so I'm well past that stage ;) | 08:55 |
| PCdude | are u telling me that u got JUJU and MAAS working together with MAAS 2.0-RC2? | 08:56 |
| magicaltrout | depends what you call working | 08:56 |
| magicaltrout | no I didn't but only because of Virtualbox sucking | 08:56 |
| magicaltrout | but yeah it was bootstrapping etc | 08:57 |
| PCdude | well, still the whole openstack with landscape shit, which needs JUJU and MAAS | 08:57 |
| PCdude | I tried the 2.0-RC2 but with an older version of JUJU, maybe the latest beta with that MAAS version might work | 08:58 |
| magicaltrout | you have to use 2.0 with 2.0 | 08:58 |
| magicaltrout | and 1.x with 1.9 | 08:58 |
| magicaltrout | but my personal opinion is that juju 2.0 is around the corner so I might as well use the beta's and put up with a few bits changing | 08:58 |
| PCdude | sorry thats what I meant, I used the JUJU beta4 or beta7 I thought with the MAAS 2.0 | 08:58 |
| magicaltrout | and its worked out for me so far | 08:58 |
| magicaltrout | on a fresh xenial juju beta15 should be current | 08:59 |
| magicaltrout | afaik | 08:59 |
| magicaltrout | and on a fresh xenial maas 2.0 is certainly the default | 08:59 |
| magicaltrout | anyway.... what bit didn't work? | 08:59 |
| PCdude | yeah, also what I have seen | 08:59 |
| PCdude | http://askubuntu.com/questions/817572/openstack-fails-to-install-caused-by-juju/818310#818310 | 08:59 |
| PCdude | in that question u can see the whole error , its a nice one haha | 09:00 |
| PCdude | I could try the older 2.0-rc2 version of MAAS, maybe that could work | 09:00 |
| magicaltrout | dunno. I was just spinning up nodes with MAAS and Juju and that was fine | 09:01 |
| magicaltrout | I've not used conjure up etc | 09:01 |
| PCdude | yeah, conjure-up is the only option for openstack with landscape | 09:02 |
| PCdude | any idea when JUJU is coming out of beta? | 09:02 |
| magicaltrout | soon | 09:02 |
| magicaltrout | its been in beta for ages | 09:02 |
| magicaltrout | so it can't be that long :P | 09:02 |
| PCdude | I hope soon too, there is a summit this week, so maybe that helps a bit | 09:03 |
| magicaltrout | well we're all off to pasadena in a couple of weeks | 09:03 |
| magicaltrout | for a juju summit | 09:03 |
| magicaltrout | its coming along, but there is a lot of tooling that needs bringing into line | 09:03 |
| PCdude | ah ok, fingers crossed :) | 09:04 |
| magicaltrout | i'll swap | 09:06 |
| magicaltrout | you fix my nagios relation | 09:06 |
| magicaltrout | and I'll sort out your open stack ;) | 09:06 |
| magicaltrout | oh and write my slides for tomorrow please | 09:06 |
| PCdude | deal! | 09:07 |
| PCdude | magicaltrout: what slides do u have to make for tomorrow? | 09:13 |
| PCdude | for a JUJU summit? or just regular work | 09:14 |
| magicaltrout | http://sched.co/7n8J | 09:15 |
| PCdude | looks good, u want me to hand ur slides before it starts ;) haha | 09:15 |
| magicaltrout | I always leave talk stuff to the last minute | 09:16 |
| magicaltrout | now I'm cramming :) | 09:16 |
| PCdude | sounds good, I am a last minute person too, most of the exams 80% was learned in the night beforehand haha | 09:18 |
| magicaltrout | i failed uni 3 times | 09:18 |
| magicaltrout | i didn't do any learning ;) | 09:18 |
| PCdude | well nothing at all is mostly not that good for ur grades, but u succeeded the 4th time? | 09:18 |
| magicaltrout | no | 09:19 |
| magicaltrout | I gave up and went to work | 09:19 |
| PCdude | I have a n00b question, how do I check if postgreSQL is running? | 09:19 |
| magicaltrout | I did fine in all the practical stuff, I just hate sitting in lectures | 09:20 |
| magicaltrout | which is ironic because I enjoy talking to people about software | 09:20 |
| PCdude | why? go to a night uni or something, or become steve jobs and make millions | 09:20 |
| magicaltrout | I don't need a degree I've been working in IT for 10 years, it pretty much covers it | 09:20 |
| PCdude | no degree problems with ur boss? | 09:21 |
| magicaltrout | infact, CS degrees a decade ago would have got you pretty much nowhere in real life | 09:21 |
| magicaltrout | I'm a self employed contractor for NASA, so I am my own boss | 09:21 |
| PCdude | CS stands for... | 09:21 |
| magicaltrout | computer science | 09:21 |
| PCdude | contractor for NASA, cool! what u do for them? | 09:21 |
| magicaltrout | Devops and data management platforms | 09:22 |
| === ant_ is now known as Guest58751 | ||
| PCdude | ah ok, u gonna tell me when they bring aliens back right? ;) | 09:22 |
| magicaltrout | ironically, I don't do much space stuff, scientific and data research projects mostly :) | 09:23 |
| PCdude | I study Electrical Engineering, but I find computer science more interesting at the moment, so yeah weird guy am I | 09:23 |
| magicaltrout | there is certainly plenty to learn in computer science, the data processing and ops platforms have all changed wildly in the last 5 or so years | 09:24 |
| PCdude | yeah true, I started with programming and have some degrees with cisco, but when I heard about openstack I jumped on the linux train and try to learn it as fast as I can | 09:26 |
| PCdude | *certifications | 09:26 |
| magicaltrout | openstack certainly has a lot of traction | 09:27 |
| magicaltrout | personally I'm more interested in Mesos type stuff | 09:27 |
| magicaltrout | to run containers across hardware | 09:27 |
| PCdude | ah ok, a bit like docker as I can see? | 09:28 |
| magicaltrout | indeed | 09:28 |
| magicaltrout | most of it is docker | 09:28 |
| magicaltrout | but the underlying stuff to manage containers is interesting, stuff like Mesos, Kubnernetes etc | 09:29 |
| magicaltrout | where you can say "here's my server pool, run my containers in the best possible way" | 09:29 |
| magicaltrout | and off it goes | 09:29 |
| PCdude | sounds cool! would u prefer that above openstack? | 09:29 |
| magicaltrout | well yeah, but each serve a different purpose | 09:30 |
| magicaltrout | like the project we're working on at NASA currently is genomic research search and discovery platforms, it makes sense to have a bunch of docker containers each with different components in | 09:30 |
| magicaltrout | and it works great because developers can deploy the identical stack locally as in production, dev or wherever | 09:31 |
| magicaltrout | but then in production we can leverage a cluster of servers and deploy docker stuff across a data center | 09:31 |
| PCdude | of course that is true, imo openstack is more widely, I mean u can add LXD or docker too, but u can also use KVM or VMware ESXI | 09:31 |
| magicaltrout | and its easier and more lightweight than running all of the openstack stuff and then containers or services over the top of that | 09:31 |
| magicaltrout | openstack might be, but i suspect kubernetes probably isn't too far behind it | 09:32 |
| PCdude | and kubernetes is more an all in one solution? | 09:33 |
| PCdude | I mean, the big con for me concerning openstack is the million packages u need | 09:33 |
| magicaltrout | its a docker management system originally by google | 09:33 |
| PCdude | free? | 09:34 |
| magicaltrout | its not entirely all in one | 09:34 |
| magicaltrout | and hard to setup, but luckily..... | 09:34 |
| magicaltrout | https://jujucharms.com/observable-kubernetes/ | 09:34 |
| PCdude | long live JUJU haha | 09:35 |
| magicaltrout | and if you're around in the afternoon lazypower hangs out here who maintains a bunch of it | 09:35 |
| magicaltrout | but that gives you kubernetes and monitoring | 09:35 |
| PCdude | cool, I am not here in the afternoon sadly... | 09:36 |
| PCdude | yeah, the problem is, I would like a little more freedom, I would like to set something up like a private cloud like the amazon aws | 09:37 |
| PCdude | and afaik, openstack is coming closest to that rn | 09:37 |
| magicaltrout | yup | 09:37 |
| magicaltrout | well I know plenty of people have had success using openstack on conjure up, so someone will know the answer to your problems | 09:38 |
| PCdude | yeah, I am currently hunting for those people, since the project itself is young ,the amount of experts is small too | 09:38 |
| magicaltrout | this is true | 09:38 |
| PCdude | I have searched the internet, but for a private version of amazon aws the only option is openstack right? I mean there are no other solutions (that is reachable for a student)? | 09:40 |
| magicaltrout | not that i'm aware of | 09:40 |
| magicaltrout | I don't pay much attention to that space, but openstack is the only main one I see on twitter etc | 09:41 |
| magicaltrout | although I might live a sheltered existence, who knows | 09:41 |
| magicaltrout | lets face it though, clouds are complex :) | 09:41 |
| magicaltrout | they will all have their own quirks | 09:41 |
| PCdude | oh so damn true! I am just a student with to much drive to get it to work, but I have put in an fair amount of time already, and nothing is working yet | 09:42 |
| magicaltrout | don't worry i know that feeling, I've been working on Kerberos based authentication for a webservice for NASA for the last week, where I have no access to the Active Directory server, stuff doesn't work and I don't know why and the Java libraries used are so old the source doesn't exist any more and I had to decompile them | 09:44 |
| magicaltrout | we've all been there ;) | 09:44 |
| magicaltrout | eventually for whatever reason stuff starts working | 09:44 |
| PCdude | java, yuck! please, lets just decide for the sake of the human race to through that ugly one of the earth | 09:45 |
| PCdude | *throw | 09:45 |
| magicaltrout | I quite like java, mostly cause its the one thing I learnt at uni | 09:46 |
| magicaltrout | but for data applications, its pretty important these days | 09:46 |
| PCdude | yeah, I understand, but all the security bugs with the application for the normal home user is insane | 09:47 |
| PCdude | the same holds for adobe flash player | 09:48 |
| magicaltrout | someone else mentioned that a few weeks ago | 09:49 |
| magicaltrout | so I showed them this | 09:49 |
| magicaltrout | http://security.stackexchange.com/questions/57646/why-do-i-hear-about-so-many-java-insecurities-are-other-languages-more-secure | 09:49 |
| magicaltrout | which makes some pretty good points | 09:49 |
| PCdude | good point | 09:53 |
| PCdude | read most of it | 09:53 |
| PCdude | yeah, I never used java in that way and only see the problems with the plugin | 09:54 |
| PCdude | that is a good point about the memory overflow though | 09:54 |
| magicaltrout | java in a browser is a terrible experience | 09:54 |
| magicaltrout | no one denies that ;) | 09:54 |
| PCdude | well honestly, I dont think java is the only problem when looking at the plugin. I think windows itself is build pretty bad too | 09:55 |
| PCdude | although they are making progress | 09:55 |
| PCdude | sandboxing should be almost compulsory these days | 09:56 |
| magicaltrout | which is basically what Snaps do for package management on Linux systems | 09:56 |
| PCdude | indeed the point I was trying to make, I convinced my parents to switch from windows to linux a few months ago | 09:57 |
| PCdude | it took 2 years, but its good now | 09:57 |
| PCdude | basically the only problem for most people is microsoft word/excel/pp | 09:58 |
| magicaltrout | true | 09:58 |
| PCdude | I am really curious though, what is gonna happen with the vulcan game engine | 10:00 |
| PCdude | some big names are making games for it now (at least that is what they say) | 10:00 |
| PCdude | the footprint of linux is smaller and could push up the FPS, maybe some of the gaming community will switch to linux | 10:00 |
| magicaltrout | probably depends if steam keep up the linux support | 10:02 |
| PCdude | true that | 10:02 |
| PCdude | how are the slides going? | 10:02 |
| PCdude | btw, what country do u live in? | 10:03 |
| magicaltrout | UK | 10:03 |
| magicaltrout | I'm on 17 of 22 | 10:03 |
| magicaltrout | they're getting there :P | 10:03 |
| magicaltrout | although I need to go to the opticians this afternoon and drive to heathrow | 10:03 |
| PCdude | wait, did we talked yesterday too? I am bad with those names so sorry if yes | 10:04 |
| magicaltrout | lol | 10:04 |
| magicaltrout | yeah we did | 10:04 |
| PCdude | sorry then haha | 10:05 |
| PCdude | what program are u using for IRC? | 10:06 |
| magicaltrout | I just irssi running on a remote server | 10:07 |
| magicaltrout | most of my life is spent inside bash terminsals, I don't see why irc would be any different | 10:07 |
| PCdude | ah ok, I use https://kiwiirc.com | 10:08 |
| PCdude | I know IRC for some time now, but have only been using it for 1 week now | 10:08 |
| magicaltrout | i have a 15 year headstart on you :P | 10:09 |
| PCdude | it is grandpa ;) | 10:09 |
| magicaltrout | i am old... | 10:10 |
| PCdude | yeah, but I could use a 15 year knowledge head start rn haha | 10:11 |
| magicaltrout | it does occasionally have some benifits being old | 10:12 |
| admcleod_ | like spelling | 10:12 |
| magicaltrout | pfft | 10:13 |
| admcleod_ | ;) | 10:13 |
| PCdude | I am dutch, so no spelling rules applied to me :D | 10:13 |
| magicaltrout | i still struggle to read teh screen admcleod_ post LASEK, luckily I have you to back m up! :P | 10:14 |
| admcleod_ | GOOD POINT | 10:14 |
| * magicaltrout ignores admcleod_ and returns to slide writing | 10:14 | |
| PCdude | read the screen? what happened? | 10:14 |
| magicaltrout | had laser eye surgery PCdude | 10:15 |
| admcleod_ | now he has laser eyes | 10:15 |
| admcleod_ | peow peow | 10:15 |
| magicaltrout | hehe | 10:15 |
| PCdude | may the force be with u | 10:17 |
| * D4RKS1D3 Hi everyone | 10:21 | |
| magicaltrout | gaa | 10:39 |
| magicaltrout | stupid nagios | 10:39 |
| magicaltrout | what on earth am i doing wrong | 10:39 |
| PCdude | good luck, with pulling out hair magicaltrout :) | 10:39 |
| PCdude | I am off to work | 10:39 |
| magicaltrout | hehe | 10:40 |
| magicaltrout | cya | 10:40 |
| PCdude | I will be here tomorrow, probably praying for an answer too haha | 10:40 |
| magicaltrout | stub: am I being a moron? to get nrpe checks registered in nagios, do you have to include monitors.yaml? | 10:53 |
| magicaltrout | i'm struggling to digest the nrpe readme | 10:53 |
| magicaltrout | or cmars | 10:53 |
| stub | I haven't used monitors.yaml, so I know it works without it. But it might be something to do with local-monitors vs. nrpe-external- | 10:54 |
| magicaltrout | sad times | 10:54 |
| magicaltrout | fair enough | 10:54 |
| stub | Best I can tell, local-monitors was intended to replace nrpe-external-master but never gained traction. I don't know who was pushing that. | 10:55 |
| stub | (this is all antique) | 10:55 |
| marcoceppi | magicaltrout stub local-monitors pre-dates nrpe-external-master, it's goal was to provide an agnostic way to describe what to monitor that could be implemented by nrpe or zabbix or any other tool | 12:27 |
| magicaltrout | yeah | 12:27 |
| marcoceppi | instead of having tens of interfaces for each monitoring tool, you just have one | 12:27 |
| magicaltrout | i was hoping for something plug and play for tomorrow | 12:28 |
| marcoceppi | now that we have interface layers, it'll probably be way easier to implement | 12:28 |
| magicaltrout | but failed | 12:28 |
| marcoceppi | but it's not anything that has traction | 12:28 |
| magicaltrout | so hacked the result | 12:28 |
| marcoceppi | FWIW, local-monitors even predates my involvement with Juju | 12:28 |
| magicaltrout | and you're bloody ancient | 12:28 |
| marcoceppi | I know >.> | 12:28 |
| magicaltrout | https://docs.google.com/presentation/d/1UGuGfU7nuJISaAEIFZ_5lm1o0o010kpdm0HMBF8jvMo/edit?usp=sharing | 12:31 |
| magicaltrout | cast your eye over them marcoceppi when you get bored | 12:31 |
| magicaltrout | i need to fill 50 minutes with slides and a demo | 12:31 |
| magicaltrout | let me know if I've missed anything obvious | 12:31 |
| marcoceppi | magicaltrout: there's a few empty slides at the end ;) but presentation flows well | 12:35 |
| magicaltrout | yeah, that bit I do know :P | 12:35 |
| magicaltrout | just doing some lunchtime hacking to try and iron out a few demo kinks | 12:35 |
| marcoceppi | cool cool, loog great so far | 12:35 |
| magicaltrout | virtualbox won't play ball and let me demo maas which is a bit sad | 12:36 |
| magicaltrout | but don't kill my instances mid presentation tomorrow please :P | 12:36 |
| marcoceppi | magicaltrout: hah, I'll keep an eye out for anything with dc/os and make sure not to reap them | 12:37 |
| marcoceppi | hey stub, had a qustion for you on layer-apt | 12:38 |
| stub | yo | 12:38 |
| marcoceppi | stub: are there plans to add something like charm.apt.does_this_package_even_exists | 12:39 |
| marcoceppi | bac Makyo charm build is fixed for your charm now | 12:39 |
| bac | marcoceppi: cool, thanks! | 12:40 |
| stub | marcoceppi: Sure. I'd like the API fat enough that all the apt stuff in charmhelpers.fetch can be deprecated. | 12:41 |
| stub | marcoceppi: But with better naming than doesthis_package_even_exist :-P | 12:41 |
| marcoceppi | stub: yeah, wasn't sure a name, just wanted to convey the concept | 12:41 |
| marcoceppi | I've got a situation in the php-fpm charm where packages exist in trusty as php5-<name> but don't in xenail as php-<name> because they're builtins now in php7 | 12:42 |
| stub | marcoceppi: The other option is delegating that to the existing apt package, but that seems to suffer from poor docs. | 12:42 |
| marcoceppi | stub: so if I ask to queue_packages and it doesn't exist will the apt layer error? | 12:42 |
| stub | marcoceppi: yes, the apt layer will error. Ignoring errors by default is one of charmhelpers.fetch's mistakes IMO | 12:43 |
| marcoceppi | stub: makes sense | 12:43 |
| marcoceppi | I'll open a bug | 12:43 |
| cmars | magicaltrout, this may help: https://paste.ubuntu.com/23111690/ | 12:43 |
| bac | marcoceppi: worked! | 12:43 |
| stub | marcoceppi: In this case, I would do 'if trusty then:' rather than 'if package in available_packages'. Because then the dead code is easily removed when trusty eols | 12:44 |
| marcoceppi | stub: well, the package name is not defined by me, it's defined by the person including the upper layer | 12:44 |
| magicaltrout | oooh | 12:44 |
| magicaltrout | very handy cmars | 12:45 |
| magicaltrout | thanks a bunch | 12:45 |
| stub | actually, it won't error. it will block with a status message informing you of the problem. | 12:45 |
| marcoceppi | stub: for example php-modules: ['mcrypt', 'mbstring', 'mysql'] | 12:45 |
| marcoceppi | php5-mbstring exists, but php-mbstring in xenail doesn't, as an example | 12:45 |
| cmars | sure thing | 12:45 |
| stub | marcoceppi: ok. So for now, I think it is 'if package not in charmhelpers.fetch.apt_cache()'. And a bug on the apt layer, since there are other use cases too. | 12:47 |
| marcoceppi | stub: cool, thanks for a path forward | 12:49 |
| magicaltrout | oooh actually that'll be killer if it works | 13:08 |
| caribou | Hello, I have a layer question : when using the basic layer & adding options: basic: packages: ['blah'], should I expect the 'blah' package to be installed when the charm is deployed ? | 13:08 |
| magicaltrout | I can nagios up my dcos master | 13:08 |
| magicaltrout | nagios up my nginx test container in the charm code | 13:08 |
| magicaltrout | yeah caribou | 13:13 |
| caribou | magicaltrout: ok, thought so; thanks! | 13:13 |
| magicaltrout | lazyPower: just the man you'll know this, if I do, juju remove-application ... | 13:14 |
| magicaltrout | can I fire a reactive hook inside that charm itself? | 13:14 |
| marcoceppi | magicaltrout: typically, the stop hook is called during remove-application | 13:14 |
| lazyPower | ^ | 13:14 |
| marcoceppi | but that's not the /only/ reason stop may fire | 13:14 |
| lazyPower | however, you also get any relationship broken/departed hooks | 13:14 |
| marcoceppi | I don't think there's a clear "tear down" hook, but maybe we should have one | 13:14 |
| magicaltrout | boo | 13:15 |
| magicaltrout | I would like a "I've removed my docker container" hook | 13:15 |
| magicaltrout | but fired from the container charm | 13:15 |
| magicaltrout | not from the dcos-master charm | 13:15 |
| marcoceppi | magicaltrout: well couldn't you do that in relation-broken ? | 13:15 |
| magicaltrout | dunno | 13:15 |
| magicaltrout | thats why i was asking | 13:15 |
| marcoceppi | that's where I'd put it | 13:15 |
| magicaltrout | do they fire within the charms that you tear down? | 13:15 |
| marcoceppi | yup | 13:16 |
| marcoceppi | juju basically "undos" what you do | 13:16 |
| magicaltrout | that'll do nicely then | 13:16 |
| marcoceppi | so if you have a relation and you remove one side of that, the relation-departed/broken hooks fire in addition to the units being torn down | 13:16 |
| marcoceppi | lazyPower: I think I found a focus that's more "smoke test" applicable | 13:18 |
| lazyPower | marcoceppi - meet you in the batcave? | 13:20 |
| marcoceppi | omw | 13:20 |
| === Anita is now known as Guest55756 | ||
| Guest55756 | Hi, getting error while deploying juju-2.0beta16 | 13:31 |
| Guest55756 | 2016-08-30 05:59:48 INFO juju.cmd supercommand.go:63 running jujud [2.0-beta16 gc go1.6.2] 2016-08-30 05:59:48 ERROR cmd supercommand.go:458 creating LXD client: Get https://10.178.209.1:8443/1.0: Unable to connect to: 10.178.209.1:8443 ERROR failed to bootstrap model: subprocess encountered error code 1 | 13:31 |
| Guest55756 | Is there any work around for this error? | 13:32 |
| Guest55756 | getting below error while deploying juju-2.0beta16 | 13:44 |
| Guest55756 | 2016-08-30 05:59:48 INFO juju.cmd supercommand.go:63 running jujud [2.0-beta16 gc go1.6.2] 2016-08-30 05:59:48 ERROR cmd supercommand.go:458 creating LXD client: Get https://10.178.209.1:8443/1.0: Unable to connect to: 10.178.209.1:8443 ERROR failed to bootstrap model: subprocess encountered error code 1 | 13:44 |
| Guest55756 | Is there any work around for this error? | 13:44 |
| babbageclunk | Guest55756: What was the command you ran? | 13:50 |
| Guest55756 | juju bootstrap local.lxd-test localhost | 13:51 |
| babbageclunk | And what does the local.lxd-test cloud look like? | 13:52 |
| babbageclunk | (I mean, how's it defined?) | 13:52 |
| babbageclunk | Guest55756: Ooh, what version of LXD are you running? | 13:53 |
| Guest55756 | LXD version - Let me check | 13:53 |
| Guest55756 | root@ptcvm3:~# lxd --version 2.0.4 root@ptcvm3:~# | 13:54 |
| Guest55756 | babbageclunk_:its 2.0.4 | 13:55 |
| === rmcall_ is now known as rmcall | ||
| Guest55756 | babbagecluck_:any work around for this error? | 14:04 |
| marcoceppi | lazyPower: so, -ginkgo.focus=Kubectl.client seems to be a decent smoke test, 32 tests including Guestbook.application | 14:05 |
| marcoceppi | basically, can I talk to the cluster via client? | 14:05 |
| Guest55756 | babbageclunk_:any work around for this error? | 14:06 |
| lazyPower | marcoceppi - its a brilliant place to start. | 14:06 |
| lazyPower | marcoceppi - let me recap whats required to do that. Just build the e2e test suite, set the ginkgo.focus= flag during run of e2e.test and thats basically it? | 14:06 |
| lazyPower | aside from the credentials dance | 14:06 |
| marcoceppi | lazyPower: just quick_release | 14:07 |
| marcoceppi | then run with a few parameters | 14:07 |
| lazyPower | allright, i'll see if i can get a job setup so we can pipeline it | 14:07 |
| babbageclunk | Guest55756: Sorry, was on the phone. | 14:07 |
| lazyPower | are you still wanting an external charm to do this validation step? | 14:07 |
| marcoceppi | lazyPower: waiting for test to finish, it's a bit long | 14:07 |
| babbageclunk | Guest55756: Are you able to launch a lxd container on that machine separately from juju? | 14:09 |
| Guest55756 | babbageclunk_:how to do that? | 14:09 |
| babbageclunk | Guest55756: Also, how hard would it be for you to upgrade to lxd 2.1? | 14:10 |
| Guest55756 | babbageclunk_:when I do apt-get install lxd, this version only comes | 14:10 |
| Guest55756 | babbageclunk_:If I do apt-get upgR | 14:11 |
| bac | marcoceppi, pmatulis: is this ready to merge? https://github.com/juju/docs/pull/1315 would like to kill those 404s | 14:11 |
| babbageclunk | Guest55756: launch a new container: `lxc launch ubuntu:` | 14:11 |
| Guest55756 | babbageclunk_:If I do apt-get upgrade lxd will it upgrade | 14:11 |
| Guest55756 | babbageclunk_:ok | 14:11 |
| Guest55756 | root@ptcvm3:~# lxc launch ubuntu: Creating vorticose-tifany Starting vorticose-tifany root@ptcvm3:~# | 14:13 |
| Guest55756 | babbageclunk_:root@ptcvm3:~# lxc launch ubuntu: Creating vorticose-tifany Starting vorticose-tifany root@ptcvm3:~# | 14:13 |
| babbageclunk | Guest55756: I think 2.1 is only in the ppa, so you'd need to do `sudo add-apt-repository ppa:ubuntu-lxc/lxd-stable` and then `sudo apt-get update` and `sudo apt-get upgrade` | 14:13 |
| Guest55756 | babbageclunk_:ok let me follow these steps and retry | 14:14 |
| Guest55756 | babbageclunk_:any other step apart from these? | 14:15 |
| babbageclunk | Guest55756: Thanks. If this works then it means we have a bug though - we should be able to work with lxd 2.0.4. | 14:15 |
| Guest55756 | babbageclunk_:will update you if its successful | 14:16 |
| babbageclunk | Guest55756: Ok, thanks. At the moment I'm not getting notifications of your replies, because they're showing up for me as babbageclunk_ - if you just say babbageclunk I'm more likely to notice if IRC isn't focused (ie. most of the time). | 14:17 |
| Guest55756 | ok :) | 14:18 |
| pmatulis | bac, i'll look at it now. i would like some feedback from your side re https://github.com/juju/docs/pull/1314 | 14:18 |
| === alexisb-afk is now known as alexisb | ||
| marcoceppi | lazyPower: so here's the output | 14:23 |
| marcoceppi | lazyPower: http://paste.ubuntu.com/23111989/ | 14:23 |
| marcoceppi | I wonder if a few of these are becuase I'm using 1.3.3 kubectl and not $master | 14:24 |
| lazyPower | doubtful | 14:24 |
| marcoceppi | \o/ | 14:24 |
| lazyPower | do we get more detailed summary? | 14:24 |
| lazyPower | or is that what we get | 14:24 |
| mattyw | mgz, ping? | 14:24 |
| marcoceppi | lazyPower: I can get a bunch of failure output | 14:25 |
| marcoceppi | but it's w3ay up in my buffer | 14:25 |
| lazyPower | ack | 14:25 |
| Guest55756 | babbageclunk:2016-08-30 13:58:37 INFO juju.cmd supercommand.go:63 running jujud [2.0-beta16 gc go1.6.2] 2016-08-30 13:58:37 ERROR cmd supercommand.go:458 creating LXD client: Get https://10.209.9.1:8443/1.0: Unable to connect to: 10.209.9.1:8443 ERROR failed to bootstrap model: subprocess encountered error code 1 | 14:25 |
| Guest55756 | babbageclunk: same error | 14:26 |
| marcoceppi | lazyPower: like this: http://paste.ubuntu.com/23111996/ | 14:26 |
| lazyPower | marcoceppi - packaging conflict has bubbled up in charm-tools pkg_resources.ContextualVersionConflict: (PyYAML 3.12 (/usr/local/lib/python2.7/dist-packages), Requirement.parse('PyYAML==3.11'), set(['jujubundlelib'])) | 14:26 |
| lazyPower | i'm seeing this in all the current implementations of charmbox, on status and in ci | 14:26 |
| marcoceppi | lazyPower: yeah, that's been fixed in master | 14:26 |
| lazyPower | ok, so bump the boxes to build from master? | 14:26 |
| mgz | mattyw: responded in other channel | 14:26 |
| marcoceppi | lazyPower: what are they building from now? | 14:27 |
| marcoceppi | pypi? | 14:27 |
| Guest55756 | babbageclunk:sorry, its still running | 14:27 |
| lazyPower | marcoceppi - apt | 14:27 |
| marcoceppi | lazyPower: that should never happen | 14:27 |
| lazyPower | https://github.com/juju-solutions/charmbox/blob/master/install-review-tools.sh#L12 | 14:28 |
| Guest55756 | babbageclunk: same error --> Reading state information... tmux is already the newest version (2.1-3build1). 0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded. Attempt 1 to download tools from https://streams.canonical.com/juju/tools/agent/2.0-beta16/juju-2.0-beta16-xenial-amd64.tgz... tools from https://streams.canonical.com/juju/tools/agent/2.0-beta16/juju-2.0-beta16-xenial-amd64.tgz downloaded: HTTP 200; time 80.146s; | 14:30 |
| Guest55756 | babbageclunk: these are the steps I followed. Please validate ---> 2009 sudo add-apt-repository ppa:ubuntu-lxc/lxd-stable 2010 sudo apt-get update 2011 sudo apt-get upgrade 2012 sudo add-apt-repository ppa:juju/devel 2013 sudo apt-get update 2014 sudo apt-get install juju-2.0 lxd 2015 lxd --version 2016 juju --version 2017 sudo lxd init 2018 #juju bootstrap lxd-test localhost 2019 juju list-controllers 2020 j | 14:31 |
| marcoceppi | lazyPower: why are you pip upgrading? | 14:31 |
| babbageclunk | Guest55756: Can you put that on paste.ubuntu.com? It's getting mangled here. | 14:32 |
| lazyPower | Tooling bump | 14:32 |
| lazyPower | we put tims test tooling bumps in both stable and devel, it required installation from source and a bump. but thats been a few weeks ago, the scenario may have changed | 14:32 |
| marcoceppi | lazyPower: well a new dep now installs PyYAML 3.12 | 14:33 |
| marcoceppi | which does not work with jujubundlelib | 14:33 |
| lazyPower | do you know which component is pulling in the new dep? | 14:33 |
| lazyPower | oh wait i see the explicit upgrade to pyyaml derp | 14:34 |
| Guest55756 | babbageclunk_: http://paste.ubuntu.com/23112045/ | 14:39 |
| babbageclunk | Guest55756: ok, looking | 14:43 |
| babbageclunk | Guest55756: When I try the same thing I get 'unknown cloud "local"' - have you got a definition for local in `juju list-clouds`? | 14:47 |
| Guest55756 | babbageclunk: let me check | 14:48 |
| babbageclunk | Guest55756: Try `juju bootstrap local lxd` instead. | 14:48 |
| Guest55756 | babbageclunk: Ok will try this. Result of previous command : root@ptcvm3:~# juju list-clouds CLOUD TYPE REGIONS aws ec2 us-east-1, us-west-1, us-west-2, eu-west-1, eu-central-1, ap-southeast-1, ap-southeast-2 ... aws-china ec2 cn-north-1 aws-gov ec2 us-gov-west-1 azure azure centralus, eastus, eastus2, northcentralus, southcentralus, westus, northeurope ... azure-ch | 14:48 |
| Guest55756 | babbageclunk: shall i paste in ubuntu.com? | 14:49 |
| babbageclunk | Guest55756: pastebin again please ;) | 14:49 |
| Guest55756 | babbageclunk: i am trying the other command | 14:50 |
| babbageclunk | Guest55756: ok | 14:50 |
| Guest55756 | babbageclunk: http://paste.ubuntu.com/23112081/ | 14:50 |
| babbageclunk | I get the same error as you're getting. | 14:52 |
| Guest55756 | ok | 14:52 |
| Guest55756 | babbageclunk: I was not getting the same error in beta15 | 14:53 |
| babbageclunk | Guest55756: Ok, so it might be a bug in beta 16 - I'm going to create a bug for it. | 14:54 |
| Guest55756 | babbageclunk_: when tried `juju bootstrap local lxd`, got the same error. Looks like a bug | 14:56 |
| Guest55756 | babbageclunk_: is there any way to take juju2.0-beta15 version? by seeting ppa:juju/ ? | 14:57 |
| Guest55756 | babbageclunk_:it take default devel version beta16 | 14:57 |
| babbageclunk | Guest55756: Try using http://askubuntu.com/questions/307/how-can-ppas-be-removed to remove the juju ppa, then remove juju and reinstall it - that should get you back to beta15 | 14:59 |
| babbageclunk | Guest55756: (There might be a more direct way than that.) | 14:59 |
| babbageclunk | Guest55756: (But I'm not sure.) | 14:59 |
| Guest55756 | babbageclunk_:ok. Thanks a lot. Will try that one | 14:59 |
| babbageclunk | Guest55756: good luck, sorry for the trouble. | 15:01 |
| Guest55756 | babbageclunk_:Thanks, np :) | 15:01 |
| marcoceppi | petevg: I had some problems using charms.unit over the weekend, mainly that it broke a whole bunch of imports when I tried Harness.patch_imports | 15:11 |
| marcoceppi | wnated to run through them witih you to see if I can figue out a way to fix it | 15:11 |
| petevg | macroceppi: I am working on an email on exactly that subject. | 15:11 |
| petevg | marcoceppi: basically, patching imports is a horrible idea. | 15:12 |
| marcoceppi | petevg: how else do we get around it? | 15:13 |
| marcoceppi | I'm using layer:apt and rely on lib/charms/apt.py from that layer | 15:13 |
| petevg | marcoceppi: I've got a PR in on the docs with some "best practices" for working around it. But basically I think that I need to scrap the harness :-( | 15:13 |
| petevg | https://github.com/juju/docs/pull/1317 | 15:13 |
| cholcombe | I have a py2 module that i need to use with reactive. Am I up a creek? | 15:27 |
| bac | pmatulis, evilnickveitch: this PR landed https://github.com/juju/docs/pull/1315 and it fixes the issue but only for 'devel'. can you update the stable tag on that branch so we don't have the 404s? i'm not sure of your criteria for moving the tag. | 15:29 |
| marcoceppi | cholcombe: kind of | 16:04 |
| bdx | is there an option that exists somewhere that will silence child layer messaging - if not this would be super useful | 16:13 |
| bdx | s/child layers/lower layers/ | 16:14 |
| bdx | also, what is the technical terminology we are using for "lower layer"|"child layer"? | 16:16 |
| marcoceppi | bdx: is it really that hurtful? I mean at the end of the day it's just telling the operator what it's doing | 16:17 |
| marcoceppi | bdx: and since you own the top layer, you set the final messages | 16:17 |
| bdx | marcoceppi: well, I only bring it up because I just demoed deploying one of our apps as a charm, people were thoroughly impressed, the only negative feedback I got was that "the ruby messaging persisting for the entirety of the life of the charm is annoying, can we change that?" | 16:24 |
| bdx | its a polish thing | 16:24 |
| marcoceppi | bdx: you totally can, in your top layer :) | 16:24 |
| bac | pmatulis, marcoceppi, evilnickveitch: automated dead link checking: https://github.com/juju/docs/pull/1318 | 16:25 |
| pmatulis | bac, i'd like to but that is the same file affected by the other bug and it has a bad link. i'd like to correct it first | 16:25 |
| bac | pmatulis: ok. if not immediately, perhaps soon. | 16:26 |
| bdx | marcoceppi: I am ... the last thing I do in my charm is -> http://paste.ubuntu.com/23112522/ | 16:27 |
| bdx | no matter what layer ruby messaging just trumps all | 16:27 |
| marcoceppi | bdx: that's a bug in ruby layer | 16:28 |
| marcoceppi | bdx: you might also want to respond to @hook('update-status') | 16:28 |
| bdx | oooh, and set status there as well | 16:28 |
| bdx | can I use @hook in a reactive layer? | 16:29 |
| marcoceppi | bdx: you can, it's discouraged, but this is a good example of where it works | 16:29 |
| marcoceppi | bdx: also, you can just have @when('prm.available') and have it always status_set... | 16:30 |
| * marcoceppi looks at ruby layer | 16:30 | |
| marcoceppi | bdx: ah, I see the problem | 16:32 |
| marcoceppi | this is a quick fix | 16:32 |
| bdx | marcoceppi: should I just not use the @when_not('prm.available') | 16:35 |
| bdx | and set_state('prm.available') | 16:35 |
| marcoceppi | bdx: no, I think that's fine | 16:35 |
| bdx | instead just ensure that it fires all the time | 16:35 |
| bdx | lol | 16:35 |
| marcoceppi | bdx: https://github.com/battlemidget/juju-layer-ruby/pull/7 | 16:38 |
| bdx | lol | 16:39 |
| bdx | YES | 16:39 |
| marcoceppi | bdx: I've also got a bunch of other fixeds to make this layer more modern I'm about to put in another PR | 16:40 |
| marcoceppi | bdx: this will have a few changes to what you're doing in your layers, as an FYI | 16:40 |
| bdx | marcoceppi: also, I've another odd little issue ... http://paste.ubuntu.com/23112614/ | 16:49 |
| bdx | no matter what I do, I can't get my charm to open ports | 16:49 |
| bdx | ha | 16:49 |
| bdx | I feel like I'm about to feel really stupid, but I've gone over everything 10x and can't seem to figure it out ... works in all my other charms implemented identically | 16:51 |
| marcoceppi | bdx: none of the charms are expsed | 16:52 |
| marcoceppi | also, lxd doesn't raelly have a firewaller | 16:52 |
| bdx | http://paste.ubuntu.com/23112623/ | 16:52 |
| marcoceppi | they should all just be "open' | 16:52 |
| bdx | on aws I get the same thing tho here .. | 16:52 |
| marcoceppi | bdx: I see the problem | 16:52 |
| bdx | even when exposed | 16:52 |
| marcoceppi | website.available is only run when you connect to the http interface | 16:52 |
| bdx | ooooooh | 16:52 |
| marcoceppi | bdx: you'll want to do that during prm.available, probably | 16:53 |
| marcoceppi | the rest of the hook looks fine, but open the port there in stead | 16:53 |
| bdx | ok | 16:54 |
| bdx | exactly what I needed | 16:54 |
| bdx | thank you thank you | 16:54 |
| marcoceppi | bdx: this is what I want to change: https://github.com/battlemidget/juju-layer-ruby/pull/8 | 16:57 |
| marcoceppi | bdx: how much of the configuration options do you use for the ruby stuff? | 16:57 |
| bdx | just 'version' | 16:58 |
| bdx | 'ruby-version' | 16:58 |
| marcoceppi | bdx: it seems that config is idempotent though, like if you change it after a deploy, it won't do anything | 16:58 |
| marcoceppi | how often do you plan on changing ruby version? | 16:59 |
| bdx | never | 16:59 |
| bdx | it should be an option then? | 16:59 |
| bdx | all of those should be | 16:59 |
| marcoceppi | bdx: yeah, that seems like a compile time option | 16:59 |
| marcoceppi | bdx: that's what I'm thinking | 16:59 |
| marcoceppi | I'll make that a separate pull request | 16:59 |
| marcoceppi | I feel bad ripping out all these chunks | 17:00 |
| bdx | anything that reduces the # of configs getting slammed in config.yaml +1 | 17:08 |
| bdx | oh I see, you haven't gotten that far | 17:12 |
| marcoceppi | bdx: eyah, not yet, I may do taht here or another PR | 17:14 |
| === frankban is now known as frankban|afk | ||
| bdx | marcoceppi: when I try to build my charm with your modernize branch -> http://paste.ubuntu.com/23112908/ | 18:13 |
| bdx | possibly I need to remove /home/bdx/allcode/charms/repo/deps ? | 18:14 |
| bdx | marcoceppi: https://github.com/battlemidget/juju-layer-ruby/issues/9 | 18:22 |
| marcoceppi | bdx: yeah, that WIP is not ready to land until charm-tools is fixed | 18:35 |
| bdx | awwwwee | 18:38 |
| cholcombe | marcoceppi, any eta on review.juju.solutions coming back up? | 18:56 |
| cholcombe | just wondering | 18:56 |
| valeech | How can I modify a charm’s bundle.yaml to deploy services to a specific maas host? I have tried changing the -to: from just 1 2 3 to hostname1.maas hostname2.maas and it fails with “placement “hostname1.maas” refers to an application not defined in this bundle” | 19:16 |
| bdx | valeech: https://github.com/jamesbeedy/os_ha_test_stack/blob/master/l3_ha_charmconf.yaml | 19:17 |
| bdx | valeech: specifically https://github.com/jamesbeedy/os_ha_test_stack/blob/master/l3_ha_charmconf.yaml#L428,L440 | 19:18 |
| bdx | thats one way | 19:18 |
| valeech | Eureka! Thanks bdx! | 19:18 |
| bdx | add constraints that are machine tags in maas | 19:18 |
| bdx | valeech: np | 19:19 |
| hatch | using tip it appears that --region is no longer supported? | 19:59 |
| === blr_ is now known as blr | ||
| === natefinch is now known as natefinch-afk | ||
| magicaltrout | marcoceppi: sometimes i'm very sad | 21:11 |
| magicaltrout | bugg@tomsdevbox:~/Projects/dcos-master$ juju deploy nagios | 21:11 |
| magicaltrout | panic: unknown channel "edge" | 21:11 |
| marcoceppi | magicaltrout: the saddest of pandas? | 21:11 |
| magicaltrout | sad because it worked this morning :P | 21:11 |
| marcoceppi | magicaltrout: yeah, a new charm store just went out, are you on beta15? | 21:11 |
| magicaltrout | yeah | 21:12 |
| marcoceppi | oh boy. | 21:12 |
| * magicaltrout 's head hits the keyboard | 21:12 | |
| magicaltrout | okay let me caveat that | 21:12 |
| magicaltrout | i have an empty beta 15 | 21:12 |
| magicaltrout | i'm not stuck to beta 15 | 21:13 |
| magicaltrout | i just need stuff to deploy | 21:13 |
| BradCrittenden | magicaltrout: please update to beta 16 | 21:13 |
| === BradCrittenden is now known as bac | ||
| magicaltrout | sage advice | 21:13 |
| magicaltrout | is that in some funky ppa? | 21:13 |
| marcoceppi | magicaltrout: ppa:juju/devel | 21:13 |
| bac | magicaltrout: we have three moving parts we tried to coordinate the release as well communication. they didn't all make it out together | 21:14 |
| bac | magicaltrout: short answer, we now have four channels and juju before beta 16 will not work | 21:14 |
| magicaltrout | no problem | 21:14 |
| magicaltrout | being locked out is something i'm used to ;) | 21:14 |
| magicaltrout | alrighty | 21:22 |
| magicaltrout | take 1 | 21:22 |
| magicaltrout | 2 | 21:22 |
| aisrael | de3b7q!0 | 21:50 |
| aisrael | gah | 21:50 |
| cmars | huh, all i see is ******** | 21:55 |
| * aisrael snickers | 21:56 | |
| magicaltrout | i see de3b7q!0 | 21:58 |
| magicaltrout | it must be some osbfucated hash | 21:58 |
| jrwren | i see hunter2 | 22:00 |
| === Guest31658 is now known as hloeung | ||
| === hloeung is now known as Guest28002 | ||
| === Guest28002 is now known as hloeung | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!