I am trying to use gcePersistentDisk as ReadOnlyMany so that my pods on multiple nodes can read the data on this disk. Following the documentation here for the same.
To create and later format the gce Persistent Disk, I have followed the instructions in the documentation here. Following this doc, I have sshed into one of the nodes and have formatted the disk. See below the complete error and also the other yaml files.
kubectl describe pods -l podName
Name: punk-fly-nodejs-deployment-5dbbd7b8b5-5cbfs
Namespace: default
Priority: 0
PriorityClassName: <none>
Node: gke-mycluster-default-pool-b1c1d316-d016/10.160.0.12
Start Time: Thu, 25 Apr 2019 23:55:38 +0530
Labels: app.kubernetes.io/instance=punk-fly
app.kubernetes.io/name=nodejs
pod-template-hash=1866836461
Annotations: kubernetes.io/limit-ranger=LimitRanger plugin set: cpu request for container nodejs
Status: Pending
IP:
Controlled By: ReplicaSet/punk-fly-nodejs-deployment-5dbbd7b8b5
Containers:
nodejs:
Container ID:
Image: rajesh12/smartserver:server
Image ID:
Port: 3002/TCP
Host Port: 0/TCP
State: Waiting
Reason: ContainerCreating
Ready: False
Restart Count: 0
Requests:
cpu: 100m
Environment:
MYSQL_HOST: mysqlservice
MYSQL_DATABASE: app
MYSQL_ROOT_PASSWORD: password
Mounts:
/usr/src/ from helm-vol (rw)
/var/run/secrets/kubernetes.io/serviceaccount from default-token-jpkzg (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
helm-vol:
Type: PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
ClaimName: my-readonly-pvc
ReadOnly: true
default-token-jpkzg:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-jpkzg
Optional: false
QoS Class: Burstable
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
---- ------ ---- ---- -------
Normal Scheduled 2m default-scheduler Successfully assigned default/punk-fly-nodejs-deployment-5dbbd7b8b5-5cbfs to gke-mycluster-default-pool-b1c1d316-d016
Normal SuccessfulAttachVolume 1m attachdetach-controller AttachVolume.Attach succeeded for volume "pvc-9c796180-677e-11e9-ad35-42010aa0000f"
Warning FailedMount 10s (x8 over 1m) kubelet, gke-mycluster-default-pool-b1c1d316-d016 MountVolume.MountDevice failed for volume "pvc-9c796180-677e-11e9-ad35-42010aa0000f" : failed to mount unformatted volume as read only
Warning FailedMount 0s kubelet, gke-mycluster-default-pool-b1c1d316-d016 Unable to mount volumes for pod "punk-fly-nodejs-deployment-5dbbd7b8b5-5cbfs_default(86293044-6787-11e9-ad35-42010aa0000f)": timeout expired waiting for volumes to attach or mount for pod "default"/"punk-fly-nodejs-deployment-5dbbd7b8b5-5cbfs". list of unmounted volumes=[helm-vol]. list of unattached volumes=[helm-vol default-token-jpkzg]
readonly_pv.yaml
apiVersion: v1
kind: PersistentVolume
metadata:
name: my-readonly-pv
spec:
storageClassName: ""
capacity:
storage: 1G
accessModes:
- ReadOnlyMany
gcePersistentDisk:
pdName: mydisk0
fsType: ext4
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: my-readonly-pvc
spec:
accessModes:
- ReadOnlyMany
resources:
requests:
storage: 1G
deployment.yaml
volumes:
- name: helm-vol
persistentVolumeClaim:
claimName: my-readonly-pvc
readOnly: true
containers:
- name: {{ .Values.app.backendName }}
image: "{{ .Values.image.repository }}:{{ .Values.image.tagServer }}"
imagePullPolicy: {{ .Values.image.pullPolicy }}
env:
- name: MYSQL_HOST
value: mysqlservice
- name: MYSQL_DATABASE
value: app
- name: MYSQL_ROOT_PASSWORD
value: password
ports:
- name: http-backend
containerPort: 3002
volumeMounts:
- name: helm-vol
mountPath: /usr/src/
I had the same error and managed to fix it using a few lines from a related article about using preexisting disks https://cloud.google.com/kubernetes-engine/docs/how-to/persistent-volumes/preexisting-pd
You need to add storageClassName
and volumeName
to your persistent volume claim:
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: pv-claim-demo
spec:
# It's necessary to specify "" as the storageClassName
# so that the default storage class won't be used, see
# https://kubernetes.io/docs/concepts/storage/persistent-volumes/#class-1
storageClassName: ""
volumeName: pv-demo
accessModes:
- ReadOnlyMany
resources:
requests:
storage: 500G
I received the same error message when trying to provision a persistent disk with an access mode of ReadWriteOnce
. What fixed the issue for me was removing the property readOnly: true
from the volumes
declaration of the Deployment spec. In the case of your deployment.yaml
file, that would be this block:
volumes:
- name: helm-vol
persistentVolumeClaim:
claimName: my-readonly-pvc
readOnly: true
Try removing that line and see if the error goes away.
It sounds like your PVC is dynamically provisioning a new volume that is not formatted with the default StorageClass
It could be that your Pod is being created in a different availability from where you have the PV provisioned. The gotcha with having multiple Pod readers for the gce volume is that the Pods always have to be in the same availability zone.
Some options:
Simply create and format the PV on the same availability zone where your node is.
When you define your PV you could specify Node Affinity to make sure it always gets assigned to a specific node.
Define a StorageClass that specifies the filesystem
kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
name: mysc
provisioner: kubernetes.io/aws-ebs
parameters:
type: gp2
fsType: ext4
And then use it in your PVC:
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: my-pvc
spec:
accessModes:
- ReadOnlyMany
resources:
requests:
storage: 1G
storageClassName: mysc
The volume will be automatically provisioned and formatted.