[09:55] How do i get the logs of a specific snap running on a juju instance (kube-apiserver in this case) [10:35] I found out eventually. (It's in syslog) [11:27] alright folks [11:27] kjackal or someone. Is there something I can prod from within an app to find out if my app is running within a charm? [11:32] scrap that found what i want [12:49] SuneK: yes these are systemd services. journalctl should get you the logs [12:51] magicaltrout: you can do a while true loop and try try to hear the fan :) [12:51] :P [12:52] lol [12:54] Where can I find the repos for the snaps used in the CDK? [12:54] SuneK: just a sec let me llok for them [12:57] SuneK: they are on the Global store, not sure what should I get you from there... [12:57] I see them when I get here: https://dashboard.snapcraft.io/snaps/ [12:58] but thats probably because i have proper permissions... [12:58] Yeah I found them there, but the developer website references an email. I would like to look into their definitions and eventually modify them [12:58] for example: https://dashboard.snapcraft.io/snaps/kube-scheduler/ [13:01] Hmm I get a 404 when I go there [13:01] Do I need special permissions? [13:02] Can you give us some more info on what you want to do? [13:02] Do you want to see the source? How these snaps are created? [13:02] I can point you to the github repos [13:02] SuneK: ^ [13:03] Exactly, I want to make the vsphere cloud provider setup easier in CDK. Github repos would be excellent [13:04] ok just a sec [13:06] SuneK: here is one: https://github.com/juju-solutions/cdk-addons [13:07] https://github.com/juju-solutions/release/tree/rye/snaps/snap [13:07] Thanks kjackal! [13:43] hi elmaciej! [13:44] hi kjackal! [13:49] elmaciej: I am eu based so you can reach me your morning hours [13:51] kjackal: Thanks a lot. I thought that maybe they will make decision today for this workshop but looks like it will take time [13:51] does greece count as europe? [13:56] lol eu for sure magicaltrout :p [13:57] jealous magicaltrout [13:57] ? [13:59] sob [13:59] i might move in [13:59] i'm very good a striking [13:59] You are always wellcomed but I though you were moving to LA [14:00] ha [14:00] trying to sell my house currently actually [14:00] the new potential buyer are fscking rubbish [14:01] you are in a bad shape if you have to do fscking [14:13] at least the UK can afford sufficient bandwidth to keep me online though kjackal [14:14] How can i re-deploy a failed envirnoment? [14:16] utking: what failed and you want to deploy a new model or do you mean to retry what failed? [14:19] magicaltrout: keep the gloves above the waist! [14:19] hehe [14:19] kjackal gives as good as he gets [14:19] this is true [14:20] I want to retry a model that failed :) [14:20] I'm deploying openstack, and it selected the wrong machine, so i have to do it again [14:21] with maas that is [14:21] you could use --to in order to select the machine [14:22] well i kind of removed all the machines, so i just need to re-deploy the model [14:22] can i use --to in the gui? [14:22] you can in a bundle I know, but I'm not sure about the ui [14:23] or do i have to that specifically on the one service (cinder) that i'm deploying? [14:23] did you destroy the model or remove the applications? [14:23] hmm [14:23] no, none of it, it took some time to set up the model the way i wanted to [14:23] when I deployed I altered my bundle to use --to [14:23] where do you do that? [14:23] so you can just add-unit myapp --to machine [14:23] `juju add-unit myapp --to machine` [14:24] ah ok :) sounds nice [14:24] even though that app is a part of a big model? [14:25] you'd have to find the bundle you were deploying and download it and edit it. Then you juju deploy that filename instead of your original bundle. Using the gui though, I don't know how you would do that, but I imagine you can... [14:25] yeah, you'd need to add all the units back that you need for the whole model [14:25] hmm, ok [14:25] so if you have 2 kubernetes-worker and 1 kubernetes-master(I know you don't with openstack, but I don't know the app names for that), you would run add-unit twice for kubernetes-worker and once for kubernetes-master [14:26] ah ok, i'll guess i'll try that then [14:26] but is there no way of just re-deploying the model i have? [14:26] I do not know what juju magic you pulled out here magicaltrout but I managed to get quassel back on! [14:26] like release the machines, and do it once more [14:27] I don't know of a way to do a deploy to a specific model, but it certainly sounds like something you could do...let me look at the docs [14:28] sure, that would be great! :) [14:28] utking: looks like you can do a juju deploy -m model_name to deploy to a specific model [14:28] i tried that [14:29] i just get no charm or bundle specified [14:29] --map-machines exists too in order to select what machines [14:29] oh it would be something like `juju deploy -m my_model openstack`, but then you'd get random machines...I think you need --map-machines to do what you want [14:30] hmm, in where in that sentence? ^^ [14:31] guess i'll read the docs some more [14:31] pretty new to juju and maas [14:36] so maybe `juju deploy -m my_model --map-machines 0=5,1=2,2=1,3=4 openstack-base` if you wanted machine 0 from the bundle to use your machine 5, which ends up with neutron-gateway, etc [14:36] I found the machine numbers and what mapped to them here: https://api.jujucharms.com/charmstore/v5/openstack-base/archive/bundle.yaml [14:37] Hi! I'm trying to relate rabbitmq and ceph-mon, when I do it it is failing as it's trying to insert rbd module which exist [14:38] should I use specific charm for rabbit on ceph based openstack installation ? I'm doing pike [14:38] ceph-relation-changed subprocess.CalledProcessError: Command '['modprobe', 'rbd']' returned non-zero exit status 1 [14:39] sorry, utking, it would be --map-machines=existing,0=5,1=2,2=1,3=4 [14:40] oh wow! thanks knobby! I'll try that out, you've been an angel :) [14:42] elmaciej: that might be an isse with the cepf relation [14:42] kjackal: it looks like, I'll try to deploy rabbit outside of the container, I have ceph-mon running without container too [14:42] wondering if I need a relation between ceph and rabbit [14:44] elmaciej: there are some restrictions when trying to modprobe inside a container [14:44] the problem is that the kernel is owned by the host operating system and it might not allow for module loading [14:45] well it should check if the module is not loaded as it is already loaded [14:49] fair point, if you open an issue on the charm code the devs will eventualy get to it [14:49] you could also submit a PR [14:49] or even "fork" the charm under your namespace [14:50] well I usually pull the charm and rebuild if it's quick fix :) [14:50] you do charm pull or do you clone the charm layer and then do a charm build? [14:51] charm pull, then fix, then charm build and install local charm [14:52] which charm are you using? [15:24] kjackal: I found proper way, none of them can be put in container and then looks good. testing now and let now if any issue appear. Thanks! === test is now known as Guest63412