| === _thumper_ is now known as thumper | ||
| === axw_ is now known as axw | ||
| === mup_ is now known as mup | ||
| uzair | How to remove charm installed completely including its container ? | 10:41 |
|---|---|---|
| [Kid] | is there a way to use hostnames with juju and MAAS to deploy certain machine numbers to certain nodes by hostname? | 15:44 |
| [Kid] | to hostname maas | 15:44 |
| rick_h | [Kid]: no, you'd have to leverage tags probably. | 15:53 |
| [Kid] | ahh okay | 15:53 |
| [Kid] | i figured that might be the case | 15:53 |
| [Kid] | i guess it wasn't really meant to get that specific | 15:54 |
| [Kid] | how about defining specific subnets for pods in k8? | 15:54 |
| [Kid] | right now i think it might just be arbitrary subnets | 15:54 |
| bdx | [Kid]: yeah `juju deploy ubuntu --to <my-maas-machine-fqdn>` | 16:08 |
| rick_h | bdx: oh can you target the fqdn | 16:13 |
| [Kid] | bdx, but can you do that in the yaml? | 16:15 |
| [Kid] | i knew about the single one offs | 16:15 |
| [Kid] | i was thinking under the machine section in the yaml | 16:15 |
| bdx | ahh my bad | 16:25 |
| bdx | not sure in that case | 16:25 |
| [Kid] | damn | 16:27 |
| [Kid] | there isn't a ton of docs around the syntax and parameters for the yaml | 16:28 |
| bdx | [Kid]: far better idea would be to tag the machines in maas and just use the tags constraints though | 16:34 |
| bdx | even if you give the machine a tag that is its hostname | 16:35 |
| [Kid] | ahh that is smart. | 16:45 |
| [Kid] | thanks | 16:45 |
| === petevg_afk is now known as petevg | ||
| === _thumper_ is now known as thumper | ||
| R_P_S | so the juju install of a kubernetes cluster creates an additional charm called kubeapi-load-balancer which is required to access the api of kubernetes. but what about apps hosted inside kubernetes? | 23:47 |
| R_P_S | do I need another load balancer (charm) to get through to the apps hosted inside kubernetes? I went through a demo, but it used GCE for it's underlying k8s, and the last part was creating a service of type "loadbalancer" | 23:48 |
| R_P_S | but this is perpetually pending and not providing an endpoint | 23:50 |
| R_P_S | it should also be noted that all instances except the kubeapi-load-balancer are in AWS private subnets, but it's not even exposing anything on the private subnet ranges | 23:51 |
| [Kid] | R_P_S, i am doing something similar. my understanding is you have to create pods on your cluster and then you expose those. | 23:54 |
| [Kid] | you have to create a YAML that describes the service you want to expose, i.e. port mainly | 23:54 |
| R_P_S | yeah, here's the yaml file from the demo that's supposed to create this endpoint | 23:59 |
| R_P_S | https://raw.githubusercontent.com/omerio/kubernetes-graphviz/master/frontend-service.yaml | 23:59 |
| R_P_S | but kubectl get service shows: frontend-service LoadBalancer 10.152.183.159 <pending> 80:32379/TCP 9m | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!