Permission error for cache volume on KinD cluster

[trisberg]

I'm building an app with riff using a 2Gi cache size. This creates the following image resource:

$ kubectl get image petclinic-application-h9vhj -oyaml
apiVersion: build.pivotal.io/v1alpha1
kind: Image
metadata:
  creationTimestamp: "2019-11-12T16:23:54Z"
  generateName: petclinic-application-
  generation: 1
  labels:
    build.projectriff.io/application: petclinic
  name: petclinic-application-h9vhj
  namespace: default
  ownerReferences:
  - apiVersion: build.projectriff.io/v1alpha1
    blockOwnerDeletion: true
    controller: true
    kind: Application
    name: petclinic
    uid: ea1b5b00-005f-45ac-8f5d-752e455b9ecc
  resourceVersion: "11822"
  selfLink: /apis/build.pivotal.io/v1alpha1/namespaces/default/images/petclinic-application-h9vhj
  uid: 68dd8d51-5ff1-49b3-95cb-1738cb08651f
spec:
  build:
    env: null
    resources: {}
  builder:
    kind: ClusterBuilder
    name: riff-application
  cacheSize: 2Gi
  failedBuildHistoryLimit: 10
  imageTaggingStrategy: BuildNumber
  serviceAccount: riff-build
  source:
    git:
      revision: master
      url: https://github.com/spring-projects/spring-petclinic.git
  successBuildHistoryLimit: 10
  tag: docker.io/trisberg/petclinic
status:
  buildCacheName: petclinic-application-h9vhj-cache
  buildCounter: 1
  conditions:
  - lastTransitionTime: "2019-11-12T16:24:23Z"
    status: "False"
    type: Ready
  - lastTransitionTime: null
    status: "True"
    type: BuilderReady
  latestBuildRef: petclinic-application-h9vhj-build-1-shpmx
  latestImage: ""
  latestStack: ""
  observedGeneration: 1

The build fails with this error:

$ kubectl logs --all-containers petclinic-application-h9vhj-build-1-shpmx-build-pod 
prepare:fetch.go:66: Successfully cloned "https://github.com/spring-projects/spring-petclinic.git" @ "32301ed5311f76ed0233ed4c08753ff9cfeaf970" in path "/workspace"
7 of 14 buildpacks participating
org.cloudfoundry.openjdk                   v1.0.48
org.cloudfoundry.buildsystem               v1.0.110
org.cloudfoundry.jvmapplication            v1.0.66
org.cloudfoundry.tomcat                    v1.1.4
org.cloudfoundry.springboot                v1.0.88
org.cloudfoundry.distzip                   v1.0.85
org.cloudfoundry.springautoreconfiguration v1.0.97
ERROR: failed to create volume cache: initializing staging directory '/cache/staging': mkdir /cache/staging: permission denied
Error from server (BadRequest): container "analyze" in pod "petclinic-application-h9vhj-build-1-shpmx-build-pod" is waiting to start: PodInitializing

Using kpack v0.0.5

The clusterbuilder I used:

$ kubectl get clusterbuilders.build.pivotal.io riff-application -oyaml
apiVersion: build.pivotal.io/v1alpha1
kind: ClusterBuilder
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"build.pivotal.io/v1alpha1","kind":"ClusterBuilder","metadata":{"annotations":{},"name":"riff-application"},"spec":{"image":"index.docker.io/cloudfoundry/cnb@sha256:4acb6bfd6c4f0cabaf7f3690e444afe51f1c7de54d51da7e63fac709c56f1c30"}}
  creationTimestamp: "2019-11-12T15:36:27Z"
  generation: 1
  name: riff-application
  resourceVersion: "1938"
  selfLink: /apis/build.pivotal.io/v1alpha1/clusterbuilders/riff-application
  uid: d859b3c6-61c3-46a5-8a0d-0de3ce8eed33
spec:
  image: index.docker.io/cloudfoundry/cnb@sha256:4acb6bfd6c4f0cabaf7f3690e444afe51f1c7de54d51da7e63fac709c56f1c30
  updatePolicy: polling
status:
  builderMetadata:
  - key: org.cloudfoundry.buildsystem
    version: v1.0.110
  - key: org.cloudfoundry.jmx
    version: v1.0.88
  - key: org.cloudfoundry.openjdk
    version: v1.0.48
  - key: org.cloudfoundry.tomcat
    version: v1.1.4
  - key: org.cloudfoundry.jvmapplication
    version: v1.0.66
  - key: org.cloudfoundry.azureapplicationinsights
    version: v1.0.90
  - key: org.cloudfoundry.go
    version: v0.0.1
  - key: org.cloudfoundry.procfile
    version: v1.0.32
  - key: org.cloudfoundry.archiveexpanding
    version: v1.0.81
  - key: org.cloudfoundry.googlestackdriver
    version: v1.0.36
  - key: org.cloudfoundry.springboot
    version: v1.0.88
  - key: org.cloudfoundry.debug
    version: v1.0.89
  - key: org.cloudfoundry.distzip
    version: v1.0.85
  - key: org.cloudfoundry.nodejs
    version: v0.0.2-RC3
  - key: org.cloudfoundry.jdbc
    version: v1.0.89
  - key: org.cloudfoundry.springautoreconfiguration
    version: v1.0.97
  - key: org.cloudfoundry.dep
    version: 0.0.51
  - key: org.cloudfoundry.go-compiler
    version: 0.0.48
  - key: org.cloudfoundry.go-mod
    version: 0.0.44
  - key: org.cloudfoundry.node-engine
    version: 0.0.49
  - key: org.cloudfoundry.npm
    version: 0.0.30
  - key: org.cloudfoundry.yarn
    version: 0.0.28
  conditions:
  - lastTransitionTime: "2019-11-12T15:36:29Z"
    status: "True"
    type: Ready
  latestImage: index.docker.io/cloudfoundry/cnb@sha256:4acb6bfd6c4f0cabaf7f3690e444afe51f1c7de54d51da7e63fac709c56f1c30
  observedGeneration: 1
  stack:
    id: io.buildpacks.stacks.bionic
    runImage: index.docker.io/cloudfoundry/run@sha256:bd9df33bff1644d567a8f74b40b2f73246325b665ae9beeef3bea7e7fa2dd3ff

[trisberg]

This seems to be specific to running on a KinD cluster. Works fine on GKE. Maybe permissions are set differently.

[trisberg]

This is most likely related to: kubernetes-sigs/kind#118