/srv/irclogs.ubuntu.com/2018/05/31/#juju.txt

wallyworldthumper: here's that PR to get the final caas things ready for 2.4. pretty please :-) https://github.com/juju/juju/pull/878401:23
* thumper looks01:23
thumperwallyworld: just a few small things01:50
wallyworldtyvm01:51
thumperI kept getting distracted while reviewing01:51
thumperemail etc.01:51
wallyworldi also have a feature test to fix01:51
wallyworldno rush i am waiting for another pr to land first01:51
wallyworldbefore i can land this01:51
thumperwallyworld: ping03:24
wallyworldyo03:25
thumperhangout again?03:25
wallyworldsure03:25
wallyworldthumper: small one? https://github.com/juju/os/pull/204:55
wallyworldneeded for caas series checking to be able to give nice errors deploying a caas charm to an iaas model04:56
veeberswallyworld: just sorting out a merge conflict and will be proposing the tomb v1 -> v2 PR05:36
veeberswallyworld: https://github.com/juju/juju/pull/8785 just noticed it could do with a commit squash, as I've merged develop in there a couple times too.05:42
wallyworldveebers: ok, looking06:09
vinowallyworld: i want u to take a look at mine as well. Made few commits06:10
wallyworldok06:10
wallyworldvino: comments left. main issue is there are mssing tests for application tools methods06:24
wallyworldplus we don't need any tests for the CLI bit as there's nothing to test really06:24
vinowallyworld: yes i am writing that now in application_test.go06:25
wallyworldgreat!06:25
wallyworldalmost there06:25
vinoYes. Wallyworld. I verified that with dump-db command as well. No IAAS and CAAS difference.06:33
vinoi will revert the chnages made in dump-db_test06:33
wallyworldty06:42
wallyworldveebers: reviewed, a mighty effort. i have aquestion, it's in the review comment06:54
veeberswallyworld: wow, I can't believe how many deletemes slipped through muy intial scan of the diff :-\07:42
=== lifeless_ is now known as lifeless
stickupkidmanadart: can I get a code review of this PR please https://github.com/juju/juju/pull/878709:32
=== salmankhan1 is now known as salmankhan
manadartstickupkid: Ja.09:51
manadartstickupkid: Reviewed.10:16
stickupkidmanadart: thanks10:17
stickupkidmanadart: i've updated the PR, we require a few hoops to jump through to now get the bridgeName - not sure if it's 100% right, but works for bionic on first try11:40
stickupkidhttps://github.com/juju/juju/pull/878711:40
manadartstickupkid: Looking.11:40
eloxbdx: elox -> erik_lonroth11:52
manadartstickupkid: Approved; couple of minor suggestions.12:04
stickupkidmanadart: nice, i'll fix those, before merging12:09
bissaIs there a way to specify a subnet when using juju to bootstrap a controller?14:00
pmatulisbissa, MAAS and AWS backing clouds support "Juju network spaces"14:07
scoobitywoops@pmatulis doesn't that require a controller to already be up?14:10
pmatulisscoobitywoops, no14:14
scoobitywoopsjuju add-space devqa 10.0.10.0/2414:15
scoobitywoopsERROR cannot connect to the API server: No controllers registered.14:15
pmatulisscoobitywoops, now i see what you mean. what *i* meant was that you can apply a space during the creation of a controller14:18
pmatulisof course, almost all commands require a controller14:18
hmlbissa: what cloud are you using?14:20
bissaaws14:20
scoobitywoopsI'm working with bissa, we're trying to build the controller in a specific vpc and subnet14:20
scoobitywoopsvpc is fine, but can't get it to stick to a subnet14:21
scoobitywoopstried --to subnet= but that didn't work14:21
chquarkhello all14:58
chquarkI need a bit of help in juju14:58
chquarkI was following this guide14:58
chquarkhttps://docs.openstack.org/project-deploy-guide/charm-deployment-guide/latest/install-maas.html14:58
chquarkwhile deploying all of the nodes on Virtualbox14:59
chquarkas a test environment14:59
chquarkuser@user:~/.local/share$ sudo juju bootstrap mymaas maas-cloud-controller --constraints "mem=1G"15:01
chquark[sudo] password for user:15:01
chquarkCreating Juju controller "maas-cloud-controller" on mymaas Looking for packaged Juju agent version 2.3.8 for amd64 No packaged binary found, preparing local Juju agent binary15:01
chquarkLaunching controller instance(s) on mymaas...ERROR failed to bootstrap model: cannot start bootstrap instance in any availability zone (default, Test)15:01
chquark^^ that is the error i get, and im at my wits end.15:02
hmlchquark: are you running with —debug ?15:02
chquarksorry Iḿ not that experienced with juju.15:03
chquarkjuju debug-log15:03
chquarkyou mean this ?15:03
hmlchquark: no15:03
hmlchquark: you can add a —debug flag (dash dash) to bootstrap for additional data15:04
chquarkgot it15:04
chquarkhttps://pastebin.com/RuGhfVrV15:07
hmlchquark: juju isn’t finding a machine to use to match the constrainsts given:  line 3615:08
hmlchquark: do you want juju to pick a specifc machine?  you can bootstrap specifying a tag as a constraint instead15:09
hmlchquark: https://docs.jujucharms.com/2.3/en/reference-constraints15:10
chquarkwell I have 2 VMś15:10
chquarkone for MAAS and another for JUJU15:11
chquarkso which machine is it trying to get ?15:11
hmlchquark:it couldn’t find a machine based on the 1G mem constraint -15:12
hmlchquark: try bootstraping without the mem piece15:12
hmlchquark: you’ll need additional VMs, with the current setup, you’ll be able to bootstrap, but there won’t be any machines available to deploy a charm15:13
chquarkI have also deployed 2 Nodes on VM, let me try it again without any contraints15:13
chquarkhttps://pastebin.com/YWEGMvZg15:16
chquarkhttps://i.stack.imgur.com/HSNAp.png15:17
hmlchquark: how much memory do the nodes commissioned by MAAS have?15:19
chquark1GB each15:20
chquarkhttps://imgur.com/a/GAQQnHt15:22
hmlchquark: juju usually looks for nodes with a min mem size of 4G… i  wonder why juju didn’t find the commissioned node when constraints were 1G15:23
hmlchquark: the min controller requirements are 2G:  https://docs.jujucharms.com/2.3/en/controllers15:24
hmlchquark: though the message you’re getting doesn’t indicated that.. juju is still trying with only 1G15:24
chquarkis this not reffering to the Juju server ?15:25
hmlchquark: what do you mean by juju server?15:25
chquarkhttps://imgur.com/tC6vITo15:27
chquarkthe Nodes have 1G ram, but MAAS and JUJU have 2G ram..15:28
hmlchquark:  in the example, node os-juju01.maas will be used as the juju controller15:29
hmlchquark: per your screen shots… is the Juju VM commissioned in maas?15:29
chquarkno Juju vm not commisioned in maas15:30
chquarkso iḿ to understand that i must add it.. let me do that right now..15:31
hmlchquark: per the example, there are 5 machines commissioned in maas for juju to use,15:31
hmlone will be the juju controller, the others are for charms to be deployed to15:32
chquarkah, crap i see where my mistake can be..15:33
hmlchquark: if you add the juju tag to node os-juju01.maas … then you can bootstrap with the tag as a constraint….15:33
chquarkyes i understand, i did not add the juju node, and had manually deployed the OS..15:34
chquarkso think i will have to redeploy that node again..15:35
hmlchquark: if you have the nodes commissioned in maas, and bootstrap juju with the maas “cloud” - juju will do the work on those nodes for you.15:38
chquark@hml, thank you for your help. I will get back to you once i have commissioned the juju in maas :)15:53
TheAbsentOneerik_lonroth: https://github.com/Ciberth/MP-appendix-a/blob/master/appendixA.md this was the guide I was working on btw!17:50
TheAbsentOneIf people are btw interested in reading my master thesis, I would love to receive feedback on wrong or unclear things. I have one more week to finish it so I'm trying to wrap things up. Ping me here if I'm here or on github/twitter: Ciberth if you want to receive the pdf!19:29
TheAbsentOneThe title is: management of polyglot persistent integrations with virtual administrators (hence the generic database use case in juju!)19:29
veebersMorning o/20:39
thumpermorning21:15
wallyworldthumper: small review? https://github.com/juju/os/pull/2 and there's a juju one that depends on it https://github.com/juju/juju/pull/878821:23
thumperack21:23
wallyworldpolishing the user experience before release21:23
* thumper nods21:23
wallyworldit used have a have a crap message21:24
veeberswallyworld: FYI I addressed your comments on PR: https://github.com/juju/juju/pull/878521:29
wallyworldveebers: yeah, just looking :-) found at least one more case of Stop() doing the err != tomb.ErrStillAlive thing21:30
wallyworldin ping batcher21:30
wallyworldi think there's 7 cases21:31
veebersah man, dropped the ball a couple times on this PR. just too eager to push it up. WIll fix21:31
veebersah I know why, faulty grep21:31
wallyworldno worries, it was a monster21:31
wallyworldveebers: also, i do think we replace donewhendying() with just <-w.Tomb.Dying21:32
wallyworldmore explicit and less overhead, no need for a func now21:32
veeberswallyworld: agreed, making it so now21:32
wallyworldawesome sauce21:32
veeberswallyworld: have pushed up the changes21:39
wallyworldveebers: looking21:46
wallyworldveebers: lgtm! also, don't forget to regenerate dependemcies.tsv before landing21:48
wallyworldand push that change21:48
wallyworldensure that tomb.v1 is missing21:48
veeberswallyworld: tomb.v1 is needed for /juju/worker.v1 :_|21:49
wallyworldoh bollocks21:49
wallyworlddo we have w worker.v221:49
wallyworldi guess that's next21:49
veeberswallyworld: indeed, has a target on it's back :-) Diff of godeps -t $(go list ./...) and deps.tsv is no changes21:50
bdxhey guys21:51
bdxsay my maas controller ip changes21:51
wallyworldthumper: we couldn't think of a different name for the OS (also called Distro in the charmstore). discussed with john and jaas guys. ideally "Kubernetes" would be the OS and we'd have "1.8" or whatever for series but the versions change too quickly for that to be managable21:51
bdxis there a way I can update the maas cloud credentials on the controller?21:52
bdxspecifically the endpoint21:52
wallyworldnot easily (yet) sadly21:52
wallyworldthe only way i know is to update mongo directly - shut down the controller agent(s) first21:52
wallyworldit's something being worked on this cycle21:53
wallyworldthumper: if we think of something better, we can change it as the charm side of things is still under development etc. fresh ideas welcome :-)21:55
bdxwallyworld: thanks21:55
wallyworldbdx: let us know if you need help etc. it's a big gap we know we need to fix21:56
bdxwallyworld: thanks, I think its going to be easier for me to just give my new maas server the ip of the old maas server21:57
wallyworldok21:57
thumperwallyworld: sure21:57
bdxsome ability to modify a cloud definition via `update-clouds` or other means would be great though21:58
bdxjust super actually21:59
bdxI'm sure more people will hit this as they upgrade21:59
bdxas maas 2.4 is bionic only21:59
bdxusers will need to add a new controller and switch to it when they upgrade22:03
wallyworldbdx: yeah, controller clouds and credentials both need to be managed better. the first thing being done is having juju better manage credentials which expire or become invalid. at the moment, things don't go well when that happens22:08
wallyworldespecially on a jaas controller with lots of models22:09
bdxI can only imagine22:13
veeberswallyworld: ah, "if err := pb.tomb.Err(); err != tomb.ErrStillAlive {" is needed for PingBatcher Stop(), that's something you pointed out to me but I'm not 100% certain why it's needed22:16
wallyworldis that the only place?22:17
veebersit seems it, will re-run tests to confirm22:18
veebersyeah looks like it. pushing fix now, PR will have a full unit test run22:19
veeberswallyworld: yay merged :-)23:04
wallyworldawesome. i would still like to look into why ping batcher needed that change23:04
veeberswallyworld: indeed, will look into it when I get a moment23:06
wallyworldveebers: we're moving to pubsub anyway soon so may not be worth the effort23:09
veeberswallyworld: ack ok23:10
wallyworldveebers: i suspect it's because someting is calling Stop() twice23:10
veebersyeah I would suspect so too23:11

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