/srv/irclogs.ubuntu.com/2016/06/02/#juju.txt

=== shawn is now known as Guest58638
=== shawn_ is now known as shawn111
=== frankban|afk is now known as frankban
admcleodbdx: hello08:28
magicaltroutf* me I'm tired. Stupid bloody kids08:30
magicaltroutdont have 'em08:30
kjackalmagicaltrout: what would you prefer human cloning or time travel?08:34
=== rogpeppe1 is now known as rogpeppe
magicaltrouthmm its a good question08:36
magicaltroutlike I worry about dying because I'll miss out on stuff. So I guess time travel falls into the same bucket08:36
magicaltroutbut at the same time, cloning only works, if I can merge at the far end ;)08:36
kjackalmagicaltrout: you could outsource your kids to your other you!08:37
magicaltrouti could simplify it and just get a nanny :P08:38
magicaltroutthat said... the mrs wont even let us get a cleaner because... and I quote "our house is too dirty"......08:38
admcleodmagicaltrout: may i make a suggestion08:56
magicaltroutsure admcleod! ;)08:57
admcleodactually i better not08:59
magicaltroutdoes it involve me replacing my kids, mrs or both? ;)08:59
admcleodyes. :D08:59
magicaltroutyeah08:59
magicaltroutits tempting08:59
admcleodi do know several people who clean their houses before their cleaner comes08:59
magicaltroutyeah its nut09:00
magicaltrouts09:00
magicaltroutlike, our house isn't bad, its just lived in, when you have 2 kids.  so why would a cleaner expect anything else?09:00
admcleodyou're not in london right? i could give you the contact for this russian cleaner... help you with all your problems...09:01
magicaltrouthehe, maybe when they get really annoying. anyway, I have 7 days off, er, i mean hard work in the states, so I don't care ;)09:02
admcleodah so thats the problem. your travelling causes you to neglect your cleaning duties. well, fair enough09:02
magicaltroutits not far off. I've been told off  quite a few times just after getting home for the state of the house, or the kids etc09:03
magicaltroutwhich is impressive as I've usually be in the US09:03
admcleodand the very next headline i see (and no, its not buzzfeed) is: 'Intelligent People Stay Up Late, Are Messy And Love To Swear'09:03
magicaltroutlol09:04
magicaltroutthe mrs goes to bed about 9:00pm and tells me off for saying bloody and knackered as they are supposedly "swear words"09:04
admcleodso, this russian cleaner i was talking about09:04
magicaltroutshe asked yesterday if at a parents evening I'd be upset with the kid if the teach said she keeps saying stuff is "bloody knackaered"09:05
admcleodhahaha09:05
magicaltroutdon't see the problem myself09:05
admcleodget her a frankie boyle bluray09:06
admcleodinfact just take her to scotland09:06
magicaltrouthehe09:06
admcleod...or like, basically outside09:09
magicaltroutyou'd think growing up in Leeds and Bradford she'd be used to that stuff09:10
magicaltroutclearly not, thats the problem with private education09:10
admcleodmaybe the problem is you're not wearing the right rugby league shirt09:10
admcleodwhile you swear09:10
magicaltrouthehe09:11
loveaHi, I'm using Juju 2.0 and MAAS 2.0. I have a MAAS server and 5 commissioned nodes (physical). I've configured a maas provider for Juju. If I juju bootstrap one of my physical nodes is utilised in maas to create the controller. That node is then no longer available for any other juju deployment. In my case that is a serious piece of kit for just a juju controller. Is there any way I can "juju deploy --to" the controller node to co-locate other charms?10:59
jcastrofginther: have you tried just deploying --to 0?12:54
rick_h_jcastro: it was a different user as I read it12:55
rick_h_jcastro: and they ahve to juju switch admin first12:55
rick_h_jcastro: to get machine 0 to deploy to12:55
jcastrooh12:57
jcastroso if you add-model foo12:57
jcastrothen you can't deploy to 0?12:57
rick_h_jcastro: no, the state server is only shown/listed in the admin model since that is the controller12:58
jcastroand a --to 0 intends to put something in the admin model?13:01
jcastroI guess I am just confused on why I can't have model foo on the same physical hardware as the admin model?13:01
rick_h_jcastro: so each model has its own machines for what's deployed into that model13:02
rick_h_rather than show 0, the bootstrap/state server on every model, it's only on the controller one13:02
jcastroahhhh, I didn't know that13:02
rick_h_especially since you can give users access to models13:02
rick_h_you don't want them adding stuff to your machine 013:02
jcastrook so basically the admin model needs a dedicated machine13:05
fgintherjcastro, I'm missing something, did you intend to ping me?13:15
jcastroI was just discussing your issue with rick_h_13:16
fgintherjcastro, sorry, do you have the bug # handy so I can have a closer look?13:19
rick_h_jcastro: it was a different person, not fginther13:19
fgintherrick_h_, ah, I'll move on then :)13:19
jcastrooh sorry, I misread my irc client when you joined right as someone was asking that13:20
kjackalkwmonroe: Hey Kevin, is this bundle going to be depricated? https://jujucharms.com/apache-processing-mapreduce/13:36
jcastromarcoceppi: got a minute? I'd like for us to unpromulgate the old mediawiki bundles13:41
jcastromramm tried one yesterday and the scalable one for sure doesn't work anymore13:42
mrammtrue that13:42
marcoceppijcastro: sure, eco-wx13:44
mrammjcastro, marcoceppi, arosales:  Is there a bug for the mysql charm's brokenness?14:01
mramm(which is what caused mediawiki-scalable to be broken)14:02
jcastronot sure14:04
jcastroiirc it was a feature we used14:04
jcastroand then was unsupported in that charm14:05
jcastroso I think it's more of the bundle not following the feature set of the charm14:05
arosalesmramm: I think we need to set the mem % on the media wiki bundle14:07
jcastrothat's a different bug14:07
arosalesAnd possible change the reference example14:07
jcastrothe master/slave config in that bundle is what isn't working14:07
jcastrothe mem% thing only affects mysql on local containers14:07
arosalesOh I guess we should define brokenness14:08
arosalesI think mem would affect small mem machined14:09
arosalesmramm: what are the symptoms?14:09
jcastrothe database doesn't come up at all in the bundle14:10
jcastrobecause it's trying to set up mysql in a master/slave configuration14:10
arosalesGotcha, and sounds like there is no bug for that14:11
arosalesShould we consider changing the example to spark/zep or observable swarm14:13
arosalesI guess swarm doesn't work in LXC14:17
mrammthe mem percentage issue seems interesting also14:41
mrammlxd/local is definitely the first use case most people touch14:42
mrammarosales: definitely we shouldn't use an example for getting started that is broken14:42
mrammthat will not get us good results14:42
marcoceppimramm: I tried fixing that, because Juju does silly things on upgrade it got nix'd14:48
mrammmarcoceppi: thanks for trying!14:53
mrammmarcoceppi: just trying things in the order that I think newbies would, and reporting issues as I run into them14:53
jcastromramm: your bug is http://pad.lv/129433414:54
jcastroarosales: the mysql master/slave thing isn't a problem in the get-started, it's the lxc memory bug that is14:56
jcastrothe master/slave feature isn't in the bundle featured on get-started14:56
jcastroall: it seems some people have tab completion in the juju betas, and some do not14:57
mrammmarcoceppi: I don't see anything about juju upgrade issues blocking a fix in the bug. Can you provide a bit more detail, what is the upgrade/juju problem with the proposed fix?14:57
jcastroif you don't have working tab completion in juju 2.0 betas, please see this bug: http://pad.lv/158840314:57
jcastromramm: it was discussed on the list, let me find it and put it in the bug14:57
mrammtab completion is working for me (at least for second level commands) on a completely fresh juju install (from the PPA), will try from Xenial repo in a min.14:59
marcoceppimramm: customers deploying mysql today under the defaults would get a change to the way the charm runs, which was viewed as unacceptable15:02
jcastromarcoceppi: is tab completion working for you?15:03
marcoceppijcastro: yes15:03
kjackalHey lazyPower, the ELK bundle under review is not pointing to logstash https://jujucharms.com/logstash/trusty/015:05
kjackalHere is the bundle http://bazaar.launchpad.net/~lazypower/charms/bundles/elk-stack/v1/view/head:/bundle.yaml15:05
lazyPowerkjackal - i think that was proposed long before any of the stack started landing15:05
lazyPoweri'll get that updated though, is that the only thing you see?15:06
kjackalI havent checked the rest of the charms15:06
kjackal let me see15:06
lazyPowerlooks like Es can use a bump15:06
lazyPoweri think thats it15:07
lazyPoweroh and kibana ;)15:07
kjackalDo you want to update them now or should we wait for the next round (next week)?15:08
lazyPowerhonestly can we push it back to next week?15:08
lazyPoweri've had a lot of stuff land and i've got a fire to put out with etcd15:09
cory_fulazyPower: Sure, kjackal can just nack the review ;)15:09
lazyPowercory_fu i'm good with that. ta15:09
mrammjcastro:  on a completely fresh xenial container -- no tab completion so not sure what's wrong on my laptop.  Haven't tried the new desktop yet.15:11
aisraeljcastro: no tabcompletion here either15:12
lazyPowersame15:12
jcastrook I'm going to ask for feedback15:19
petevgHi, magicaltrout. I'm taking a look at the saikuanalytics-enterprise charm in the review queue. I'm following the README, and the action that generates a license seems to execute, but I get an error when logging in ("could not find license"). Are there any steps to fetching the license that are missing from the README?15:29
petevg(Looking at logs, I see the error about not being able to find the license, but not necessarily any errors related to fetching it in the first place ...)15:30
magicaltroutgood find on the tab completetion jcastro !15:59
magicaltroutit makes me sad everytime I press tab15:59
magicaltroutpetevg: don't worry about it for now, I'm gonna get 3.9 into the review queue and prod someone to do it in a quicker fashion as it will have much improved juju/big data integration once my drill stuff is finished16:00
petevgmagicaltrout: Sounds good. I look forward to seeing the new charm :-)16:08
magicaltroutyou and me both! :P16:10
magicaltroutalthough 4.0 later in the year will be the stuff. Nice new React.js responsive UI, pluggable datasources and UI elements etc16:11
magicaltroutthats like the holy grail ;)16:11
lazyPowerbig data with a front end that doesn't look like java?16:12
lazyPower\o/16:12
magicaltrouthehe. Well our UI is starting to get a bit old, its all Backbone and HTML, but web tech moves so fast its a pain to keep up16:13
magicaltroutbut we also wanted to build a new setup built upon server and UI OSGI modules, so people can write OSGI compliant 3rd party plugins and just have the app register them16:13
magicaltroutwhich will make extendability and updates much nicer16:14
lazyPowernot bad obama16:14
magicaltroutbecause you'll be able to do OSGI updates from a maven repo over the wire16:14
magicaltroutrather than downloading a new 700mb distro16:14
lazyPowerThats a win16:22
roadmrhey folks, is there a "cron" charm? say I want an existing unit with e.g. apache to run a cron entry to generate some static content. Do I have to write a custom charm for that or is there something I can just deploy --to the unit with the command to run as a "juju set" parameter?16:28
lazyPowerno but that sounds like a great subordinate service. the ability to provide it with a crontab and have it bolt that into the unit16:34
lazyPowermy initial grep of the charm store yields no results that look relevant. i think you've found some golden territory roadmr16:35
roadmr\o/ /o\16:36
roadmrthanks lazyPower16:36
=== frankban is now known as frankban|afk
=== barry` is now known as barry_
=== barry_ is now known as barry
arosalesjcastro: I have a bug opened for tab completion let me find the number17:28
arosaleshttps://bugs.launchpad.net/bugs/158201817:29
mupBug #1582018: Tab completion for file system paths and in general for stock Ubuntu cloud image <bash-completion> <charms> <juju-core:Triaged> <https://launchpad.net/bugs/1582018>17:29
LiftedKiltturns out juju add-unit {charm} -n {#} --to lxc:{machine} creates one lxc container and then tells maas it needs physical machines for the rest18:05
natefinchfwereade: think I figured it out. Or rather, dpb1_ figured it out - the error message looks like it's coming from the local machine agent, but since it's actually an error in state, it must be on the controller, and passed back to the machine agent, which logs it without indicating it's logging an error from an API call to the controller :/18:22
dpb1_let's hope so18:22
alexisbnatefinch, ug18:27
alexisbthough good on the progress18:27
=== redir is now known as redir_lunch
=== redir_lunch is now known as redi
=== redi is now known as redir
kjackalkwmonroe: Hey Kevin, is this bundle going to be depricated? https://jujucharms.com/apache-processing-mapreduce/19:39
kjackalcory_fu you might know as well ^19:40
cory_fuYes, it is19:41
cory_fuIn favor of hadoop-processing19:41
kjackalhadoop processin19:41
kjackallet me see19:41
cory_fuThat's the Bigtop bundle19:41
kjackalit the apache-processing-mapreduce going to be removed from the store?19:42
kjackalPanagiotis has a reference on the bundle on his paper and I just want to make sure his link points to something19:43
kjackalcory_fu?19:44
cory_fuHow is it being referenced?  cs:bundle/apache-processing-mapreduce?19:46
cory_fuI think we were considering unpromulgating it, but leaving it in the ~bigdata-charmers namespace19:46
kjackalhm... I will try to rephrase that part so that it references our bigdata landing page then....19:48
kjackalI think this is the safest....19:48
kjackaljog: I am with the bigdata team. What is the customer impact of the "Hadoop install hooks fail behind restricted network" ?20:36
kjackaljog: the issue is already in the "doing" state, I wonder who is deploying the bundle in restricted environments. DO we have a customer?20:37
mbruzekcmars: Is there a way to list the terms I have already accepted?21:01
=== natefinch is now known as natefinch-afk
cmarsmbruzek, juju list-agreements21:04
mbruzek[]21:04
mbruzekcmars: then how was I able to deploy a local charm with terms for loren-ipsum/1 ?21:05
cmarsmbruzek, locally deployed charms do not check terms21:05
cmarsmbruzek, terms are required by the charmstore21:05
mbruzekOh I see21:05
jogkjackal, any enterprise installation is most certainly going to have their data center behind a firewall21:09
kjackaljog: understood, I was hoping for a specific enterprise name where we failed tp deploy21:11
kjackaljog: how did you discover the fault in that bundle?21:11
jogkjackal, even Canonical's own data center is an example... the system test hardware the cdo-qa team uses is on a restricted IS provided network21:12
kjackalI see, thank you. Again reviewing all our bundles to filter out all similar isues in in doing "state"21:15
mbruzekcmars: Do you have a minute?21:53
cmarsmbruzek, sure21:53
mbruzekcmars: Is it possible to make a resource "optional" ?21:53
mbruzekWe are trying to charm push a charm with one optional resource and it is complaining that we don't give it any resources21:54
cmarsmbruzek, hmm that's a good question. I'm not too familiar with the finer details there. i can definitely see the usefulness in that21:55
mbruzekcmars: That is right you implemented terms21:55
cmarsalexisb, who's around today who would know about mbruzek's resources question? ^^21:55
mbruzekcmars: Who would know... ahh21:56
alexisbwallyworld or natefinch-afk can answer mbruzek q21:56
mbruzekwallyworld: We are trying to "charm publish" a charm and the error message is telling us we must have both resources referenced when we publish.  The last resource is optional21:57
cmarsmbruzek, i suppose you *could* do some hacky thing like rc=/dev/null but that's less than ideal21:58
cmarswell, on publish, it's not too bad to do that21:58
mbruzekcmars: Yes that is what kwmonroe and I were thinking, but less than ideal.21:58
cmarsit's unixy21:58
wallyworldmbruzek: it may be that at the moment, all resources do need to be defined21:58
wallyworldfrom memory, i don't think optional resources were considered in the requirements21:59
mbruzekwallyworld: thank you22:00
cmarsmbruzek, i can see both sides to this.. on one hand, you might have forgotten to specify the resource when you publish the charm22:00
wallyworldcan you raise a bug?22:00
cmarsmbruzek, on the other, you might want it to be optional22:00
cmarsfor the first case, i'd appreciate the error, for the second, i wouldn't22:00
mbruzekcmars: Yes I see both sides of the coin.22:00
cmarsmaybe the resource ought to declare default behavior?22:01
cmarsor that it's optional22:01
mbruzekcmars: wallyworld: well we coded the charm to have the second resource optional22:01
mbruzekwe worked around this problem by pushing zero length files, but that feels weird.22:02
cmarsmbruzek, true, but the flip side is, you don't get empty files when you forget resources. i think we'd need an optional: attribute in the metdata22:02
mbruzekcmars: wallyworld: I will raise a bug and we can think about the best way to implement this22:03
wallyworldsounds good22:03
cmarsmbruzek, cool, thanks. just thinking out loud..22:03
mbruzekwallyworld: https://bugs.launchpad.net/juju-core/+bug/158855522:50
mupBug #1588555: Can resources be optional? <juju-core:New> <https://launchpad.net/bugs/1588555>22:50
wallyworldmbruzek: ty22:50
rick_h_mbruzek: answered22:52
rick_h_wallyworld: ^22:53
mbruzekthanks rick_h_22:54
magicaltroutwhy is the mongodb charm running such an ancient version of mongo? operational reasons or just no one got around to updating it?23:00
lazyPowermagicaltrout - thats the case yes :(23:18
lazyPoweri'm the listed maintainer but i have not touched mongodb in quite some time23:18
lazyPowerand it was in a state of mostly working in smaller cluster but constantly a headache at scale. With reactive it should turn into a  much nicer charm to maintain, and we can model the individual components of the charm as top layers instead of a single catch-all multi-role charm23:19
lazyPowerwhich in all honestly makes the charm code abysmal to read through23:19
magicaltroutno problem lazyPower i just noticed it was well old23:19
magicaltrouti'll see if I can get around to it at some point23:19
lazyPoweryeah, i really want someone to adopt that one and give it a good home23:19
lazyPoweri've been a crappy parent of that charm23:19
lazyPowersorry :(23:20
magicaltroutI don't have much of a use for it, but clearly there is a lot of mongo use in the real world23:20
magicaltroutbe ashame to put people off because its 2.423:20
lazyPowerour resident mongo master has been busy hacking on snappy related stuff23:20
magicaltroutI was saying to SaMnCo today as well, we(everyone where possible) should make a bigger effort to make the charms centos/rhel compatible as well23:22
magicaltroutfor the enterprisey folk. It was a sticky issue with the guys we were having lunch with23:22
lazyPowermatt was just looking at a MP for charmhelpers from cloudbase introducing the centos abstractions for charmhelpers.core.host23:22
lazyPowerits getting close but has some wrap up fixes needed23:22
magicaltroutcool23:22
lazyPoweri really want to try to start the migration process from charmhelpers into a trim charms.host lib and thinks of that nature23:23
lazyPowerextracting them and making it all travisy and publish frequently if the tests pass23:23
lazyPowerbut thats a way bigger project than i'm making it out to be23:23
lazyPowerand i have a habit of wanting to take on the world every time i get a second23:23
magicaltroutaye23:24
magicaltrouti have that problem23:24
lazyPowerwhat can i say, we <3 this stuff23:24
* lazyPower wanders off to brew some tea23:24
magicaltrouthttp://spicule.co.uk/2016/06/02/apache-drilll-juju.html brain dump on Apache Drill charm23:24
magicaltroutstarted it at 8am and finished at midnight with a 12 hour gap in between23:27
lazyPowernot bad23:27
lazyPowerimpressive actually23:28
magicaltroutto many l's in drill23:29
magicaltroutarse23:29
magicaltroutthat was an 8am error ironically23:29
magicaltroutlack of morning coffee23:29
lazyPowerWhat did hemmingway say? write intoxicated, edit sober.23:36
magicaltroutaye,he had a very valid point23:37
aisraellazyPower: indeed23:47
lazyPower;)23:47
lazyPowerdidnt know you were lurking around over there aisrael o/23:48
lazyPowerall unpacked?23:48

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!