Minikube: latest storage-provisioner: cannot get resource "endpoints" in API group "" in the namespace "kube-system"

[tolusha]

Not sure, but it seems related to https://github.com/kubernetes/minikube/pull/6511/files

The exact command to reproduce the issue:

minikube start
kubectl create namespace test
kubectl apply -f pvc.yaml

The full output of the command that failed:

storage provisioner logs:

E0206 10:46:39.353949       1 leaderelection.go:331] error retrieving resource lock kube-system/k8s.io-minikube-hostpath: endpoints "k8s.io-minikube-hostpath" is forbidden: User "system:serviceaccount:kube-system:storage-provisioner" cannot get resource "endpoints" in API group "" in the namespace "kube-system

The output of the minikube logs command:

The operating system version:

minikube version
minikube version: v1.5.2
commit: 792dbf92a1de583fcee76f8791cff12e0c9440ad-dirty

cat /etc/os-release 
NAME="Linux Mint"
VERSION="19.3 (Tricia)"

pvc.yaml:

kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: test
  namespace: test
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 1Gi

[dimara]

We, Arrikto, also bumped into this with MiniKF, which uses Minikube v1.2.0.

It seems that the storage-provisioner image was retagged yesterday. Specifically:
OLD:

gcr.io/k8s-minikube/storage-provisioner                            v1.8.1                                     sha256:088daa9fcbccf04c3f415d77d5a6360d2803922190b675cb7fc88a9d2d91985a   4689081edb10        2 years ago         80.8MB

NEW:

gcr.io/k8s-minikube/storage-provisioner                            v1.8.1                                     sha256:341d2667af45c57709befaf7524dc1649a3b025d63a03ee6d48013574043c93d   c74c2f8485c3        6 hours ago         30.7MB

This is very likely to cause problems and break existing software/deployments, as it has now. Pushing different images with the same tags should be avoided, right?

[dimara]

I see that PR #6156 changes the storage-provisioner code. @tstromberg @nanikjava shouldn't the STORAGE_PROVISIONER_TAG variable be changed after this?

[marlowl]

Got a related problem I guess: kubectl describe pvc is giving the following message: waiting for a volume to be created, either by external provisioner "k8s.io/minikube-hostpath" or manually created by system administrator. PVC stuck at pending. Running Minikube v1.7.1

[tstromberg]

Likely related to #6496

Since the bug is affecting older versions, it sounds like we rolled out a buggy provisioner image. I'll see if I can back it out. In the future we need to make sure to not repush to specific revisions.

[tstromberg]

@dimara thanks for the hashes!

I ran:

docker tag gcr.io/k8s-minikube/storage-provisioner@sha256:088daa9fcbccf04c3f415d77d5a6360d2803922190b675cb7fc88a9d2d91985a gcr.io/k8s-minikube/storage-provisioner:v1.8.1
docker push gcr.io/k8s-minikube/storage-provisioner:v1.8.1

Please let me know if this has helped in your environment.

[VlasLive]

@dimara thanks for the hashes!

I ran:

docker tag gcr.io/k8s-minikube/storage-provisioner@sha256:088daa9fcbccf04c3f415d77d5a6360d2803922190b675cb7fc88a9d2d91985a gcr.io/k8s-minikube/storage-provisioner:v1.8.1
docker push gcr.io/k8s-minikube/storage-provisioner:v1.8.1

Please let me know if this has helped in your environment.

Helped in our env.

[tolusha]

It works on my side.

[dimara]

@tstromberg It works now. Thanks!

[tstromberg]

Thanks!

[tstromberg]

This should no longer affect users, but I'm leaving this open until we address the root cause.

Process fix: https://github.com/kubernetes/minikube/pull/6526
Root cause fix TBD

[djmarcin]

I was broken by this for a while before I figured out that in order to fix I have to completely purge my ~/.minikube directory so that it will fetch the corrected v1.8.1. Would it be good to publish a v1.8.2 so that anyone who already has the broken v1.8.1 image gets the update?

[ChenRoth]

is there any plan on having a newer tag for the fix? the alternative is to hack minikube into working :(

[bodom0015]

Possibly a duplicate of #3129, which appears to still be present in v1.8.1

[afbjorklund]

Pushing a v1.8.2 sounds like a good fix for this (the bad binary).

[ianfoo]

Another user report in case it helps anyone:

I'd read this thread a number of times and still couldn't figure out how to get past having PVCs stuck in Pending. I had the right hash on the provisioner image, based on what's shown above. However, when I checked pods in namespace kube-system I saw storage-provisioner in Terminating state, stuck for some reason. I force deleted this pod (with --force --grace-period 0) and restarted minikube (v1.8.1) and the storage provisioner pod came up successfully, and my PVCs were bound to newly created PVs. Unblock'd! 🙌🏼

[limbuu]

@tstromberg
Thanks it works now.

I was having same issue in minikube version: v1.8.1. I have used virtualbox as VM driver.

$ minikube version
minikube version: v1.8.1
commit: cbda04cf6bbe65e987ae52bb393c10099ab62014

$ kubectl apply -f mysql-pvc.yaml 
persistentvolumeclaim/mysql-data-disk created

$ kubectl get pvc
NAME              STATUS    VOLUME   CAPACITY   ACCESS MODES   STORAGECLASS   AGE
mysql-data-disk   Pending                                      standard       14m

 kubectl describe pvc mysql-data-disk
Name:          mysql-data-disk
Namespace:     default
StorageClass:  standard
Status:        Pending
Volume:        
Labels:        <none>
Annotations:   kubectl.kubernetes.io/last-applied-configuration:
                 {"apiVersion":"v1","kind":"PersistentVolumeClaim","metadata":{"annotations":{},"name":"mysql-data-disk","namespace":"default"},"spec":{"ac...
               volume.beta.kubernetes.io/storage-provisioner: k8s.io/minikube-hostpath
Finalizers:    [kubernetes.io/pvc-protection]
Capacity:      
Access Modes:  
VolumeMode:    Filesystem
Mounted By:    mysql-deployment-655cdd8d98-rqbrf
Events:
  Type    Reason                Age                  From                         Message
  ----    ------                ----                 ----                         -------
  Normal  ExternalProvisioning  2m (x26 over 8m15s)  persistentvolume-controller  waiting for a volume to be created, either by external provisioner "k8s.io/minikube-hostpath" or manually created by system administrator

When I dig deeper, I found that my storage-provisioner was enabled but its pod was evicted.

$ minikube addons list
|-----------------------------|----------|--------------|
|         ADDON NAME          | PROFILE  |    STATUS    |
|-----------------------------|----------|--------------|
| dashboard                   | minikube | disabled     |
| default-storageclass        | minikube | enabled ✅   |
| efk                         | minikube | disabled     |
| freshpod                    | minikube | disabled     |
| gvisor                      | minikube | disabled     |
| helm-tiller                 | minikube | disabled     |
| ingress                     | minikube | disabled     |
| ingress-dns                 | minikube | disabled     |
| istio                       | minikube | disabled     |
| istio-provisioner           | minikube | disabled     |
| logviewer                   | minikube | disabled     |
| metrics-server              | minikube | disabled     |
| nvidia-driver-installer     | minikube | disabled     |
| nvidia-gpu-device-plugin    | minikube | disabled     |
| registry                    | minikube | disabled     |
| registry-creds              | minikube | disabled     |
| storage-provisioner         | minikube | enabled ✅   |
| storage-provisioner-gluster | minikube | disabled     |
|-----------------------------|----------|--------------|

$ kubectl get pods -n kube-system
NAME                          READY   STATUS    RESTARTS   AGE
coredns-6955765f44-fcnsw      1/1     Running   26         28d
coredns-6955765f44-tjlgt      1/1     Running   25         28d
etcd-m01                      1/1     Running   17         24d
kube-apiserver-m01            1/1     Running   2864       24d
kube-controller-manager-m01   1/1     Running   1481       28d
kube-proxy-wwm2q              1/1     Running   25         28d
kube-scheduler-m01            1/1     Running   1481       28d
storage-provisioner           0/1     Evicted   0          28d

Now its working,

$ docker pull gcr.io/k8s-minikube/storage-provisioner:v1.8.1
v1.8.1: Pulling from k8s-minikube/storage-provisioner
Digest: sha256:088daa9fcbccf04c3f415d77d5a6360d2803922190b675cb7fc88a9d2d91985a
Status: Image is up to date for gcr.io/k8s-minikube/storage-provisioner:v1.8.1
gcr.io/k8s-minikube/storage-provisioner:v1.8.1

$ docker images
REPOSITORY                                TAG                 IMAGE ID            CREATED             SIZE
k8s.gcr.io/kube-proxy                     v1.17.3             ae853e93800d        2 months ago        116MB
k8s.gcr.io/kube-apiserver                 v1.17.3             90d27391b780        2 months ago        171MB
k8s.gcr.io/kube-controller-manager        v1.17.3             b0f1517c1f4b        2 months ago        161MB
k8s.gcr.io/kube-scheduler                 v1.17.3             d109c0821a2b        2 months ago        94.4MB
<none>                                    <none>              af341ccd2df8        3 months ago        5.56MB
kubernetesui/dashboard                    v2.0.0-beta8        eb51a3597525        4 months ago        90.8MB
k8s.gcr.io/coredns                        1.6.5               70f311871ae1        5 months ago        41.6MB
k8s.gcr.io/etcd                           3.4.3-0             303ce5db0e90        6 months ago        288MB
kubernetesui/metrics-scraper              v1.0.2              3b08661dc379        6 months ago        40.1MB
gluster/glusterfile-provisioner           latest              1fd2d1999179        23 months ago       233MB
k8s.gcr.io/pause                          3.1                 da86e6ba6ca1        2 years ago         742kB
gcr.io/k8s-minikube/storage-provisioner   v1.8.1              4689081edb10        2 years ago         80.8MB

$ kubectl get pods -n kube-system
NAME                          READY   STATUS    RESTARTS   AGE
coredns-6955765f44-fcnsw      1/1     Running   26         28d
coredns-6955765f44-tjlgt      1/1     Running   25         28d
etcd-m01                      1/1     Running   17         24d
kube-apiserver-m01            1/1     Running   2864       24d
kube-controller-manager-m01   1/1     Running   1481       28d
kube-proxy-wwm2q              1/1     Running   25         28d
kube-scheduler-m01            1/1     Running   1481       28d
storage-provisioner           1/1     Running   0          114s

$ kubectl apply -f mysql-pvc.yaml 
persistentvolumeclaim/mysql-data-disk unchanged

$ kubectl get pvc
NAME              STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS   AGE
mysql-data-disk   Bound    pvc-4e8b1ff1-9080-449c-afbd-deca623d4b0c   100Mi      RWO            standard       42m

[limbuu]

@tstromberg
Although pvc and pv are created, another problem keeps the pod in pending state.

$ kubectl logs -f storage-provisioner -n kube-system
E0424 08:40:25.014427       1 controller.go:682] Error watching for provisioning success, can't provision for claim "default/mysql-data-disk": events is forbidden: User "system:serviceaccount:kube-system:storage-provisioner" cannot list resource "events" in API group "" in the namespace "default"
E0424 08:59:56.106645       1 controller.go:682] Error watching for provisioning success, can't provision for claim "default/mysql-data-disk": events is forbidden: User "system:serviceaccount:kube-system:storage-provisioner" cannot list resource "events" in API group "" in the namespace "default"

Inside mysql pod,

$ kubectl get pods
NAME                                READY   STATUS    RESTARTS   AGE
mysql-deployment-655cdd8d98-8cfv9   0/1     Pending   0          10s

$ kubectl describe pod mysql-deployment-655cdd8d98-8cfv9
Name:           mysql-deployment-655cdd8d98-8cfv9
Namespace:      default
Priority:       0
Node:           <none>
Labels:         app=mysql
                pod-template-hash=655cdd8d98
Annotations:    <none>
Status:         Pending
IP:             
IPs:            <none>
Controlled By:  ReplicaSet/mysql-deployment-655cdd8d98
Containers:
  mysql:
    Image:      mysql:5.7
    Port:       3306/TCP
    Host Port:  0/TCP
    Environment:
      MYSQL_ROOT_PASSWORD:  <set to the key 'MYSQL_ROOT_PASSWORD' in secret 'mysql-secret'>  Optional: false
      MYSQL_DATABASE:       <set to the key 'MYSQL_DATABASE' in secret 'mysql-secret'>       Optional: false
      MYSQL_USER:           <set to the key 'MYSQL_USER' in secret 'mysql-secret'>           Optional: false
      MYSQL_PASSWORD:       <set to the key 'MYSQL_PASSWORD' in secret 'mysql-secret'>       Optional: false
    Mounts:
      /var/lib/mysql from mysql-data (rw,path="mysql")
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-bcgp4 (ro)
Conditions:
  Type           Status
  PodScheduled   False 
Volumes:
  mysql-data:
    Type:       PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
    ClaimName:  mysql-data-disk
    ReadOnly:   false
  default-token-bcgp4:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-bcgp4
    Optional:    false
QoS Class:       BestEffort
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type     Reason            Age        From               Message
  ----     ------            ----       ----               -------
  Warning  FailedScheduling  <unknown>  default-scheduler  persistentvolumeclaim "mysql-data-disk" not found
  Warning  FailedScheduling  <unknown>  default-scheduler  persistentvolumeclaim "mysql-data-disk" not found
  Warning  FailedScheduling  <unknown>  default-scheduler  0/1 nodes are available: 1 node(s) had taints that the pod didn't tolerate.

FYI, I have not added any taints or node selector or node affinity anti-affinity rules.
And, after couple hours of running, storage-provisioner pod get evicted again.

$ kubectl get pods -n kube-system
NAME                          READY   STATUS    RESTARTS   AGE
coredns-6955765f44-fcnsw      1/1     Running   26         28d
coredns-6955765f44-tjlgt      1/1     Running   25         28d
etcd-m01                      1/1     Running   17         24d
kube-apiserver-m01            1/1     Running   2864       24d
kube-controller-manager-m01   1/1     Running   1481       28d
kube-proxy-wwm2q              1/1     Running   25         28d
kube-scheduler-m01            1/1     Running   1481       28d
storage-provisioner           0/1     Evicted   0          163m

[limbuu]

The pod was evicted by low disk space, removing unused images and the intermediate images (none tags) and stopped containers fixed the issue for me.

$ docker rm $(docker ps -a -q)

$ docker rmi $(docker images -f dangling=true -q )

[mostafa-eltaher]

This worked for me and fixed the issue on minikube 1.12.1
https://github.com/kubernetes/minikube/issues/8414#issuecomment-667422690

[sharifelgamal]

This should be fixed with the newest version of storage-provisioner. I'm going to close this issue for now but if someone continues to see issues, open a new bug or comment here.