pmatulis | i rebooted a machine and status shows 'hook failed: "leader-settings-changed"'. how do i get out of that? | 01:24 |
---|---|---|
anastasiamac | pmatulis: 'juju resolved <unit-name>', see for 'juju help resolved' for more info | 01:32 |
anastasiamac | pmatulis: u might need to do it for all units on that machine.. altho we r hoping u have a unit/machine | 01:32 |
pmatulis | anastasiamac, hi! awesome. i did it for three units on the machine and everything looks fine now | 01:38 |
anastasiamac | pmatulis: \o/ | 01:39 |
pmatulis | anastasiamac, btw, what is an elegant way to simulate a downed unit? nova-compute/1 to be precise | 01:41 |
anastasiamac | pmatulis: i dont know specifics for nova-compute but when i need to have a 'downed' unit, i stop it's machine | 01:42 |
pmatulis | anastasiamac, yeah, that's what got me into trouble :) i neglected the fact that this is a hyperconverged openstack node | 01:44 |
anastasiamac | pmatulis: :) | 01:44 |
pmatulis | i guess i'll go monkey around with the processes in the machine | 01:45 |
anastasiamac | k | 01:45 |
thumper | pmatulis: the best way is to ssh into the machine and stop the unit agent | 02:12 |
pmatulis | thumper, and that will stop the corresponding "service" (e.g. nova-compute)? | 02:42 |
pmatulis | (same as 'systemctl stop nova-compute.service ?) | 02:43 |
thumper | pmatulis: no | 02:52 |
thumper | if you are trying to replciate a workload down | 02:52 |
thumper | then you need to take the workload down, not the agent | 02:52 |
thumper | unless you are trying to replicate a machine down | 02:52 |
pmatulis | thumper, right, that's what i thought | 02:53 |
wallyworld | kelvinliu: lgtm! a few small things before landing. let's get it in and make progress | 04:43 |
kelvinliu | wallyworld: just back from lunch, thx for review, | 05:24 |
wallyworld | no worries | 05:24 |
kelvinli_ | hi wallyworld saw some of the comments are not different with the spec, got time HO to discuss further? | 06:16 |
wallyworld | sure | 06:17 |
wallyworld | kelvinli_: forgot to ask - with ken's external-ip issue - is it just sufficient for us to assign a user supplied external ip value passed in at bootstrap time to the correspondoing "external-ips" controller service attribute | 06:35 |
kelvinli_ | wallyworld: im not sure, need to take a look further | 06:35 |
wallyworld | ok, next week :-) | 06:35 |
kelvinli_ | yep | 06:35 |
manadart | Trivial review: https://github.com/juju/juju/pull/10579 | 07:22 |
elox | /msg NickServ identify 1ircpassword | 07:39 |
elox | fantastic passwordchange? | 07:40 |
manadart | achilleasa: I was talking to rick_h about the network/space remodelling work last week and he mentioned you would be in the slot to move on to this soon. | 08:12 |
manadart | This is worth a read, as it is something we are looking into as part of the work: https://discourse.jujucharms.com/t/multiple-space-bindings-per-endpoint/1999 | 08:13 |
achilleasa | manadart: thanks for the link! | 08:17 |
stickupkid | achilleasa, if the series isn't valid and we don't ask the user to use force, can they still use force? | 10:11 |
achilleasa | stickupkid: I guess they could but it would still fail right? Could it be a valid series that the client doesn't know of yet? | 10:13 |
stickupkid | achilleasa, so if the client doesn't know about it, we don't either, so in that instance we would need a new release... using force wouldn't help either, as no binaries... but it seems very total. | 10:14 |
achilleasa | stickupkid: I guess we could leave it as-is then. It's highly unlikely that people will try to bootstrap with an invalid series name to begin with, right? | 10:17 |
=== jose is now known as Guest4255 | ||
achilleasa | stickupkid: I pushed a commit to my packaging PR which addresses the review comments. As per John's suggestion I will extract the FromURL method and move it to my upcoming juju PR | 10:45 |
stickupkid | achilleasa, i'll swap you then https://github.com/juju/os/pull/12 | 10:56 |
achilleasa | stickupkid: approved | 11:01 |
stickupkid | achilleasa, ta | 11:06 |
stickupkid | achilleasa, i approved yours as well | 11:06 |
stickupkid | good spot about users in urls | 11:06 |
stickupkid | that was a disaster waiting to happen | 11:06 |
achilleasa | stickupkid: that's why we should never log errors :D | 11:14 |
achilleasa | stickupkid: I removed the URL bits. Can you do a final check before I merge? | 11:15 |
stickupkid | achilleasa, yeap, happy with that | 11:18 |
manadart | stickupkid: I responded to you comment. in my patch. Take a look when you've the time. | 12:39 |
rick_h | stickupkid: manadart made a suggestion for the wording. Let me know what you think. | 14:00 |
manadart | rick_h: Works for me; will mod. | 14:00 |
stickupkid | rick_h, yarp, much better | 14:02 |
hml | manadart: i updated the comments for items you had questions in my pr. pls take a look and see if they make more sense | 14:12 |
manadart | hml: Thanks. All looks good. | 14:21 |
magicaltrout | hello fine people | 15:30 |
magicaltrout | i need to bootstrap a kubernetes cloud | 15:30 |
magicaltrout | and its been a while and i'm stuck | 15:30 |
magicaltrout | rick_h: wake up! ;) | 15:33 |
stickupkid | magicaltrout, think he's out atm, where you stuck? | 15:35 |
magicaltrout | just trying to figure out the bootstrap docs stickupkid | 15:36 |
magicaltrout | https://paste.ubuntu.com/p/MHvTVQVyyd/ | 15:36 |
magicaltrout | so anyway | 15:36 |
magicaltrout | i have a k8s cluster with not much in it, few namespaces and a couple of pods and i need to write some k8s charms | 15:37 |
magicaltrout | so i'm trying to bootstrap it | 15:37 |
magicaltrout | works using kubectl on the same box as juju | 15:37 |
magicaltrout | but i get that, but i can't find any docs telling me what blanks I should be filling in there | 15:37 |
magicaltrout | for kubernetes spun up from juju, what is the cloud name/type/region blah | 15:38 |
magicaltrout | I just get told its wrong when I guess | 15:38 |
stickupkid | magicaltrout, is it a local one? | 15:38 |
stickupkid | magicaltrout, "juju add-k8s kubernetes --local" would work in that case | 15:39 |
magicaltrout | well thats the other thing, i saw local in the help.. i have zero clue what local refers to | 15:39 |
magicaltrout | local to what? | 15:39 |
magicaltrout | oh like "If you just | 15:39 |
magicaltrout | want to update the local cache and not a running controller, use | 15:39 |
magicaltrout | the --local option." | 15:39 |
stickupkid | magicaltrout, yeah | 15:39 |
magicaltrout | well I have a controller, is this not the generic controller? | 15:39 |
magicaltrout | like its on a box which already has a juju connected to an openstack cloud... | 15:40 |
magicaltrout | don't i just use that controller? | 15:41 |
magicaltrout | its not clear | 15:41 |
pmatulis | magicaltrout, generally you add a cloud to your local client | 15:44 |
magicaltrout | consider me well confused... we always use controllers and now i'm being told to add stuff to a local client... | 15:45 |
tvansteenburgh | magicaltrout: it sounds like you're trying to do the bootstrap step before you do the `juju add-k8s` step | 15:47 |
magicaltrout | no i'm running the juju add-k8s step tvansteenburgh | 15:47 |
magicaltrout | infact | 15:48 |
magicaltrout | local gives me the same error | 15:48 |
magicaltrout | i still dont' know what to type! :) | 15:48 |
tvansteenburgh | And I don't know what you've already typed! :) | 15:48 |
magicaltrout | https://paste.ubuntu.com/p/MXkSKsmQgH/ | 15:48 |
magicaltrout | that was my last guess cause stickupkid told me to do --local | 15:48 |
pmatulis | magicaltrout, just curious, did you look over any of the documentation? maybe that stuff needs improving | 15:49 |
magicaltrout | on that box, juju status, shows my bootstrapped controller and kubectl get namespaces runs fine | 15:49 |
magicaltrout | pmatulis: https://discourse.jujucharms.com/t/using-kubernetes-with-juju/1090 | 15:49 |
magicaltrout | i have this page open | 15:49 |
magicaltrout | i have run juju add-k8s --help | 15:50 |
magicaltrout | and i'm absolutely non the wiser | 15:50 |
tvansteenburgh | magicaltrout: Pipe your kubeconfig to add-k8s | 15:50 |
magicaltrout | i've tried a few different methods of getting kubectl in | 15:52 |
magicaltrout | but i also know its reading it cause i messed the file up | 15:52 |
magicaltrout | and it failed with another error message | 15:52 |
magicaltrout | https://asciinema.org/a/X2LWuBEpolBYYvoT79JFT8zwm | 15:52 |
magicaltrout | i mean, it all seems to be working, so i'm clearly missing something dumb but its not obvious in the docs | 15:53 |
magicaltrout | i mean | 16:07 |
magicaltrout | if i wanted to fill out the cloud and region | 16:07 |
magicaltrout | what the hell goes in them? | 16:07 |
magicaltrout | I do see any examples anywhere | 16:07 |
achilleasa | stickupkid: I am trying to find a place in the juju code-base to add the "get snap store assertions" helper. Any ideas? "snap/assertions" would be great but "snap" is used for the snapcraft bits | 16:12 |
achilleasa | (there is also a service/snap which doesn't seem right) | 16:12 |
stickupkid | achilleasa, do they reference any of the juju/juju code base, if not, then core/snap/assertions? | 16:12 |
stickupkid | although core should be renamed to pkg or internal :) | 16:13 |
achilleasa | stickupkid: no, it's just the bit that I removed from the juju/packaging PR. Ok, I will put it there for now... | 16:14 |
magicaltrout | jeez | 16:15 |
magicaltrout | it actually added something | 16:15 |
magicaltrout | so tvansteenburgh to get it to add the cloud i had to run | 16:17 |
magicaltrout | juju add-k8s k8s-test-cloud --debug --region openstack/RegionOne --storage openstack-standard | 16:18 |
pmatulis | https://discourse.jujucharms.com/t/tutorial-installing-kubernetes-with-cdk-and-using-auto-configured-storage/1469#heading--adding-the-cluster-to-juju | 16:18 |
pmatulis | magicaltrout, ^^^ | 16:18 |
pmatulis | also: https://bugs.launchpad.net/juju/+bug/1830949 | 16:18 |
mup | Bug #1830949: [k8s] add-k8s command has ambigious UX <usability> <juju:Fix Committed by anastasia-macmood> <https://launchpad.net/bugs/1830949> | 16:18 |
magicaltrout | ah yeah that tutorial has the region in as well pmatulis yeah, i picked that up a few minutes ago from another tutorial | 16:20 |
magicaltrout | trying to bootstrap now thanks | 16:21 |
rick_h | magicaltrout: geeze, had to take the dog to the vet. Sorry :P | 16:25 |
rick_h | magicaltrout: you get going? | 16:25 |
magicaltrout | no probs rick_h | 16:26 |
magicaltrout | the thing seems to be trying to bootstrap | 16:26 |
rick_h | magicaltrout: that's a good thing | 16:26 |
magicaltrout | that add-k8s command is a mindfuck | 16:27 |
rick_h | lol, a little bit | 16:27 |
rick_h | magicaltrout: going to get some food, but feedback/etc in discourse is helpful for sure. | 16:28 |
magicaltrout | I think the issue pmatulis linked to in launchpad captures the problem pretty well | 16:29 |
rick_h | I'm hoping we can get to a point that add-cloud/add-k8s are pretty much the same walk through of stuff vs the different worlds they have now. | 16:29 |
rick_h | magicaltrout: yea | 16:29 |
magicaltrout | but if i hit more I'll bring it up. I need to get some k8s charms written for Druid | 16:29 |
magicaltrout | so I'm sure i'll hit some more fun | 16:29 |
magicaltrout | epic, bootstrap came to life | 16:34 |
magicaltrout | thanks folks | 16:34 |
magicaltrout | random k8s charm question | 20:53 |
magicaltrout | username: %(docker_image_username)s | 20:53 |
magicaltrout | password: %(docker_image_password)s | 20:53 |
magicaltrout | what are they when they're at home? | 20:53 |
magicaltrout | like, don't you push the docker image as a resource, so what does it relate to? | 20:53 |
magicaltrout | image_info.registry_path.. | 20:57 |
magicaltrout | etc. is that just some juju thing then? | 20:57 |
rick_h | magicaltrout: hmm, not seeing that in the spec definition. Is that something image specific? | 21:16 |
magicaltrout | na its in all of them... it looks like its how juju authenticates with the juju docker repo I guess | 21:16 |
rick_h | oh maybe | 21:17 |
magicaltrout | well when I say all of them | 21:18 |
magicaltrout | I mean mediawiki and mariadb | 21:18 |
magicaltrout | my sample pool :P | 21:18 |
magicaltrout | okay basic charm works | 22:16 |
magicaltrout | thats pretty cool | 22:16 |
magicaltrout | deployment: | 22:16 |
magicaltrout | type: stateless | stateful | 22:16 |
magicaltrout | service: cluster | loadbalancer | external | 22:16 |
magicaltrout | anyone know where that goes? to get a loadbalancer IP? | 22:16 |
magicaltrout | it seems to suggest it goes in metadata.yaml | 22:16 |
magicaltrout | "Charm metadata syntax looks like this" | 22:16 |
magicaltrout | but if I stick it in there charm build tells me to get lost | 22:17 |
magicaltrout | proof: E: Unknown root metadata field (deployment) | 22:18 |
magicaltrout | ah thats just a proof thing | 22:27 |
magicaltrout | it still deploys and does what it claims | 22:27 |
magicaltrout | good stuff | 22:27 |
=== lborda is now known as lborda_afk |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!