invalid charm name

[papertigers]

Trying to use juju to deploy kubernetes via http://kubernetes.io/docs/getting-started-guides/juju/ and I am hit with the following:

$ cluster/kube-up.sh
... Starting cluster using provider: juju
... calling verify-prereqs
... calling kube-up
build: Processing layer: layer:basic
build: Processing layer: layer:docker
build: Processing layer: layer:flannel
build: Processing layer: layer:tls
build: Processing layer: kubernetes
build: Processing interface: etcd
build: Processing interface: tls
error: invalid charm name "/home/ubuntu/kubernetes/cluster/juju/bundles/local.yaml"

[mbruzek]

Hello @papertigers sorry that you encountered this problem.

We are working on a new getting started document. It is getting reviewed as I type: https://github.com/chuckbutler/kubernetes.github.io/pull/2

This not the right repository to open issues with. Did you get the link from the old document?

[arosales]

@papertigers if you looking for a recent K8 bundle I would suggest: https://jujucharms.com/u/containers/auditable-kubernetes/bundle/2 https://jujucharms.com/u/containers/kubernetes-core/bundle/0 Unless @mbruzek or @chuckbutler have a more recent bundle they would recommend.

If your interested in Containers and Juju, in general, I would take a look at: http://containers.juju.solutions/

[lazypower]

Greetings @papertigers

I'm extremely happy to see you're taking a look at our Kubernetes charms. However, it appears that you've run into a papercut while we continue working through promoting our latest work.

The actual issue here is that the instructions you are following are targeted at juju 1.25 (current stable)

In preparation for Juju 2.0, we've adjusted the tooling, as well as the documentation (still cycling, no pull request yet) to focus on the new tooling and get ahead of the curve, as the 2.0 launch date is scheduled for April. This leaves us with a chicken/egg problem since our code lives in teh upstream repository.

One thing to note, is unless you plan on hacking on the Kubernetes source, you're better to just use the charm store revisions of the charms, as they will always be stable, current, and our recommended production deployment formations. What you get out of the K8s repository is actually a smoke test formation with a single etcd node, and 2 kubernetes nodes.

@arosales linked to two bundles, the first is a bundle that ships with log aggregation and visualization, the second is the core of Kubernetes - and probably where you want to start. it ships with 3 etcd nodes (HA formation), and 2 kubernetes nodes. You can scale the Kubernetes cluster up and down, and it has support for log shipping via logspout or elastic beats.

This is easily deployable via juju-quickstart, or juju deploy if you are running juju 2.0

juju-quickstart cs:~containers/bundle/kubernetes-core

I hope this helps, and I apologize in advance for the confusion while we continue making improvements to our area in the cluster directory. And thanks a bunch for filing the bug, we appreciate the feedback!

[papertigers]

Hey @chuckbutler it got me further this time.
I followed https://github.com/mbruzek/kubernetes.github.io/blob/mbruzek-juju/docs/getting-started-guides/juju.md

But maybe I should just be doing juju deploy, like you state above?

[Units]
ID           WORKLOAD-STATUS JUJU-STATUS VERSION   MACHINE PORTS PUBLIC-ADDRESS MESSAGE
etcd/0       active          idle        2.0-beta3 0             37.153.110.89  Etcd leader running
kubernetes/0 error           idle        2.0-beta3 1             37.153.108.78  hook failed: "install"
kubernetes/1 error           idle        2.0-beta3 2             37.153.111.167 hook failed: "install"

I deleted those VM's on my side but it appears juju doesn't seem to realize that everything is gone (sorry I am new to juju). Is there a way to cleanup the env easily?

[papertigers]

Never mind it looks like things are in .local/share/juju and clearing things out was enough to get me out of the issue. I am sure there is a better way to clean things up?

Also let me attempt to deploy again.