kubeflow / website

Kubeflow Website
https://www.kubeflow.org
Creative Commons Attribution 4.0 International
150 stars 777 forks source link

Test and fix the guide to Microk8s/Multipass for Kubeflow installation #1113

Closed shouryaotta closed 4 years ago

shouryaotta commented 5 years ago

I need to use the latest mlpipelines released in 0.6.2 and am using a desktop to try it out. I have installed microk8s using the instructions at https://www.kubeflow.org/docs/other-guides/virtual-dev/getting-started-multipass/

I used the latest executable for 0.6.2 with the following instructions.

# The following command is optional, to make kfctl binary easier to use.
export PATH=$PATH:<path to kfctl in your kubeflow installation>

# Initialize a kubeflow app:
export KFAPP=<your choice of application directory name> (ensure this is lowercase)
kfctl init ${KFAPP}

# Generate and deploy the app:
cd ${KFAPP}
kfctl generate all -V
kfctl apply k8s -V

I get the following error message :

INFO[0039] creating ClusterRoleBinding/ml-pipeline-scheduledworkflow  filename="kustomize/kustomize.go:447"
INFO[0039] creating Deployment/ml-pipeline-scheduledworkflow  filename="kustomize/kustomize.go:447"
Error: couldn't apply KfApp:  (kubeflow.error): Code 500 with message: kfApp Apply failed for kustomize:  (kubeflow.error): Code 500 with message: couldn't create default profile from &{{Profile kubeflow.org/v1alpha1} {kubeflow-anonymous      0 0001-01-01 00:00:00 +0000 UTC <nil> <nil> map[] map[] [] nil [] } {{User  anonymous }} { }} Error: no matches for kind "Profile" in version "kubeflow.org/v1alpha1"

Usage:
  kfctl apply [all(=default)|k8s|platform] [flags]

Flags:
  -h, --help      help for apply
  -V, --verbose   verbose output default is false

couldn't apply KfApp:  (kubeflow.error): Code 500 with message: kfApp Apply failed for kustomize:  (kubeflow.error): Code 500 with message: couldn't create default profile from &{{Profile kubeflow.org/v1alpha1} {kubeflow-anonymous      0 0001-01-01 00:00:00 +0000 UTC <nil> <nil> map[] map[] [] nil [] } {{User  anonymous }} { }} Error: no matches for kind "Profile" in version "kubeflow.org/v1alpha1"

Earlier the templates were using argoproj.io. Does something else need to be updated for the new profile to work ?

issue-label-bot[bot] commented 5 years ago

Issue-Label Bot is automatically applying the label kind/feature to this issue, with a confidence of 0.64. Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback!

Links: app homepage, dashboard and code for this bot.

sarahmaddox commented 4 years ago

See further info in the following two issues:

I'll close those issues and we'll track the work on this issue, since it's the first one raised.

jtfogarty commented 4 years ago

/area docs /priority p2

sarahmaddox commented 4 years ago

@ammarn911 is tackling this issue

sarahmaddox commented 4 years ago

Status: PR https://github.com/kubeflow/website/pull/1469 has added an alpha banner to the Microk8s installation doc. There are still some questions to resolve - see comments in the PR, so I'll leave this issue open.

kbhawkey commented 4 years ago

A couple of thoughts about deploying kubeflow on workstations/appliances:

There are a number of places in the kubeflow docs that start to outline how to deploy kubeflow in a single node cluster (and workstation/appliance) and finish in another set of instructions. This can be confusing.

I tried to install kubeflow on a microk8s cluster and Ubuntu; I followed the instructions from https://master.kubeflow.org/docs/other-guides/virtual-dev/getting-started-multipass/.

On this page, it may help to clarify the version of kubeflow that is available to install (and what components are in that package). Or, can you specify the version by a channel flag? I installed Kubernetes 1.15 instead of 1.17 (by way of the channel flag).

How is Ambassador deployed or needed (routing)? Is there an explanation about Ambassador in the kubeflow docs?

During a second microk8s+kubeflow installation/deployment, I used the kubeflow manifests (kfctl_k8s_istio.0.7.1.yaml) and deployed into microk8s. The deployment was successful, though, my cluster became unresponsive and I had difficulty removing kubeflow from the cluster. I read through the following pages: https://master.kubeflow.org/docs/other-guides/kustomize/ https://master.kubeflow.org/docs/started/getting-started/

I would be interested in deploying a smaller set of kubeflow components that support a couple of samples/demos (deploy the examples with the prerequisite components).

sarahmaddox commented 4 years ago

Thanks @kbhawkey that's super useful feedback.

/cc @jlewi /cc @ammarn911

jlewi commented 4 years ago

@ammarn911 I think we discussed this at the community meeting. I think you said you have testing in place?

The other Canonical engineer mentioned that there is testing in this repo https://github.com/juju-solutions/bundle-kubeflow/

Is that covering the current installation guide

I think the updated docs are now visible here https://master.kubeflow.org/docs/other-guides/virtual-dev/getting-started-multipass/

What is the follow on work to #1469?

jlewi commented 4 years ago

See https://github.com/kubeflow/website/pull/1609#issuecomment-583255010

Do we need to add that to the instructions?

Also it looks like there is a reference to Ambassador in the guide https://github.com/kubeflow/website/blob/master/content/docs/other-guides/virtual-dev/getting-started-multipass.md#2-enable-kubeflow

That needs to be updated.

/cc @knkski

jlewi commented 4 years ago

Has this been fixed can we close it?

sarahmaddox commented 4 years ago

I think the Microk8s/Multipass doc still needs some work. For example, it still refers to Ambassador: https://www.kubeflow.org/docs/started/workstation/getting-started-multipass/

stale[bot] commented 4 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

issue-label-bot[bot] commented 4 years ago

Issue-Label Bot is automatically applying the labels:

Label Probability
area/kfctl 0.51

Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback! Links: app homepage, dashboard and code for this bot.