Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[K3S] Can't start cache-deployer-deployment #4138

Closed
haibingzhao opened this issue Jul 2, 2020 · 8 comments
Closed

[K3S] Can't start cache-deployer-deployment #4138

haibingzhao opened this issue Jul 2, 2020 · 8 comments
Labels
area/backend help wanted The community is welcome to contribute. kind/feature status/triaged Whether the issue has been explicitly triaged

Comments

@haibingzhao
Copy link
Contributor

What steps did you take:

# - ../gcp/inverse-proxy
  • Deployed the pipeline locally with K3D:
kubectl apply -k manifests/kustomize/cluster-scoped-resources
kubectl apply -k manifests/kustomize/env/dev/
  • k3d and k3s version is:
➜  ~ k3d --version
k3d version v3.0.0-rc.5
k3s version v1.18.4-k3s1 (default)

What happened:

The module cache-deployer-deployment can't start because of:

+ kubectl certificate approve cache-server.kubeflow
No resources found
Error from server (Forbidden): certificatesigningrequests.certificates.k8s.io "cache-server.kubeflow" is forbidden: user not permitted to approve requests with signerName "kubernetes.io/legacy-unknown"

What did you expect to happen:

Environment:

Localhost environment on K3D:

k3d create kubeflow -w 3
k3d get kubeconfig kubeflow --switch

How did you deploy Kubeflow Pipelines (KFP)?

kubectl apply -k manifests/kustomize/cluster-scoped-resources
kubectl apply -k manifests/kustomize/env/dev/

KFP version: 1.0.0-rc.2

KFP SDK version: not use

Anything else you would like to add:

Related to this recent change in Kube: https://github.com/kubernetes/kubernetes/pull/86933/files
I think the cache-deployer clusterrole needs to be tweaked(manifests/kustomize/base/cache-deployer/cluster-scoped/cache-deployer-clusterrole.yaml):

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  labels:
    app: kubeflow-pipelines-cache-deployer-clusterrole
  name: kubeflow-pipelines-cache-deployer-clusterrole
rules:
- apiGroups:
  - certificates.k8s.io
  resources:
  - certificatesigningrequests
  - certificatesigningrequests/approval
  verbs:
  - create
  - delete
  - get
  - update
- apiGroups:
  - admissionregistration.k8s.io
  resources:
  - mutatingwebhookconfigurations
  verbs:
  - create
  - get
- apiGroups:
  - certificates.k8s.io
  resources:
  - signers
  resourceNames:
  - kubernetes.io/legacy-unknown
  verbs:
  - approve

add 'approve' (verb) 'signers' (resource) for 'certificates.k8s.io' (apiGroup), it works now. report this bug.

/kind bug
/area backend

@Bobgy
Copy link
Contributor

Bobgy commented Jul 2, 2020

/assign @Ark-kun
@haibingzhao Can you explain what kubernetes version you are using?

@Bobgy
Copy link
Contributor

Bobgy commented Jul 2, 2020

Ohhh, is it k3s version v1.18.4-k3s1 (default)?

@haibingzhao
Copy link
Contributor Author

Ohhh, is it k3s version v1.18.4-k3s1 (default)?

yes

@Bobgy Bobgy added the status/triaged Whether the issue has been explicitly triaged label Jul 3, 2020
@Bobgy
Copy link
Contributor

Bobgy commented Jul 3, 2020

I think we have only been testing on Kubernetes 1.14 which is the stable channel in GCP.

PR welcomed in making it compatible with 1.18

@Bobgy Bobgy added the help wanted The community is welcome to contribute. label Jul 7, 2020
k8s-ci-robot pushed a commit that referenced this issue Jul 22, 2020
…s-cache-deployer-role. Fixes #4138 (#4246)

* certificat approval

* update pr

* update pr

* update pr

* update pr

* remove list&watch
@alfsuse
Copy link

alfsuse commented Aug 3, 2020

Hi, @Bobgy I see this issue is still open, I've tested the changes merged in #4246 on k3s and on Kind with K8s 1.18, I've also tested it on a "standard" K8s (not DinD) and I can confirm it works. Do you need still/more help on this? If so how can I help you?

@Bobgy
Copy link
Contributor

Bobgy commented Aug 3, 2020

Thank you!
/close

@k8s-ci-robot
Copy link
Contributor

@Bobgy: Closing this issue.

In response to this:

Thank you!
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

Ark-kun pushed a commit to Ark-kun/pipelines that referenced this issue Aug 17, 2020
…s-cache-deployer-role. Fixes kubeflow#4138 (kubeflow#4246)

* certificat approval

* update pr

* update pr

* update pr

* update pr

* remove list&watch
k8s-ci-robot pushed a commit that referenced this issue Aug 18, 2020
…s-cache-deployer-role. Fixes #4138 (#4246) (#4383)

* certificat approval

* update pr

* update pr

* update pr

* update pr

* remove list&watch

Co-authored-by: Eugene Fedorenko <eugene.fedor@gmail.com>
Jeffwan pushed a commit to Jeffwan/pipelines that referenced this issue Dec 9, 2020
…s-cache-deployer-role. Fixes kubeflow#4138 (kubeflow#4246) (kubeflow#4383)

* certificat approval

* update pr

* update pr

* update pr

* update pr

* remove list&watch

Co-authored-by: Eugene Fedorenko <eugene.fedor@gmail.com>
@ReggieCarey
Copy link

Confirmed that the bug still exists in 1.2 but the above repair is valid on Kubernetes 1.18.9.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
area/backend help wanted The community is welcome to contribute. kind/feature status/triaged Whether the issue has been explicitly triaged
Projects
None yet
Development

No branches or pull requests

6 participants