Kubernetes pod pending when a new volume is attached (EKS)

↘锁芯ラ 提交于 2019-12-05 19:56:15

问题


Let me describe my scenario:

TL;DR

When I create a deployment on Kubernetes with 1 attached volume, everything works perfectly. When I create the same deployment, but with a second volume attached (total: 2 volumes), the pod gets stuck on "Pending" with errors:

pod has unbound PersistentVolumeClaims (repeated 2 times)
0/2 nodes are available: 2 node(s) had no available volume zone.

Already checked that the volumes are created in the correct availability zones.

Detailed description

I have a cluster set up using Amazon EKS, with 2 nodes. I have the following default storage class:

kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
  name: gp2
  annotations:
    storageclass.kubernetes.io/is-default-class: "true"
provisioner: kubernetes.io/aws-ebs
parameters:
  type: gp2
reclaimPolicy: Retain
mountOptions:
  - debug

And I have a mongodb deployment which needs two volumes, one mounted on /data/db folder, and the other mounted in some random directory I need. Here is an minimal yaml used to create the three components (I commented some lines on purpose):

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  namespace: my-project
  creationTimestamp: null
  labels:
    io.kompose.service: my-project-db-claim0
  name: my-project-db-claim0
spec:
  accessModes:
  - ReadWriteOnce
  resources:
    requests:
      storage: 5Gi
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  namespace: my-project
  creationTimestamp: null
  labels:
    io.kompose.service: my-project-db-claim1
  name: my-project-db-claim1
spec:
  accessModes:
  - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  namespace: my-project
  name: my-project-db
spec:
  replicas: 1
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        name: my-db
    spec:
      containers:
        - name: my-project-db-container
          image: mongo
          imagePullPolicy: Always
          resources: {}
          volumeMounts:
          - mountPath: /my_dir
            name: my-project-db-claim0
          # - mountPath: /data/db
          #   name: my-project-db-claim1
          ports:
            - containerPort: 27017
      restartPolicy: Always
      volumes:
      - name: my-project-db-claim0
        persistentVolumeClaim:
          claimName: my-project-db-claim0
      # - name: my-project-db-claim1
      #   persistentVolumeClaim:
      #     claimName: my-project-db-claim1

That yaml works perfectly. The output for the volumes is:

$ kubectl describe pv

Name:            pvc-307b755a-039e-11e9-b78d-0a68bcb24bc6
Labels:          failure-domain.beta.kubernetes.io/region=us-east-1
                failure-domain.beta.kubernetes.io/zone=us-east-1c
Annotations:     kubernetes.io/createdby: aws-ebs-dynamic-provisioner
                pv.kubernetes.io/bound-by-controller: yes
                pv.kubernetes.io/provisioned-by: kubernetes.io/aws-ebs
Finalizers:      [kubernetes.io/pv-protection]
StorageClass:    gp2
Status:          Bound
Claim:           my-project/my-project-db-claim0
Reclaim Policy:  Delete
Access Modes:    RWO
Capacity:        5Gi
Node Affinity:   <none>
Message:        
Source:
    Type:       AWSElasticBlockStore (a Persistent Disk resource in AWS)
    VolumeID:   aws://us-east-1c/vol-xxxxx
    FSType:     ext4
    Partition:  0
    ReadOnly:   false
Events:         <none>


Name:            pvc-308d8979-039e-11e9-b78d-0a68bcb24bc6
Labels:          failure-domain.beta.kubernetes.io/region=us-east-1
                failure-domain.beta.kubernetes.io/zone=us-east-1b
Annotations:     kubernetes.io/createdby: aws-ebs-dynamic-provisioner
                pv.kubernetes.io/bound-by-controller: yes
                pv.kubernetes.io/provisioned-by: kubernetes.io/aws-ebs
Finalizers:      [kubernetes.io/pv-protection]
StorageClass:    gp2
Status:          Bound
Claim:           my-project/my-project-db-claim1
Reclaim Policy:  Delete
Access Modes:    RWO
Capacity:        10Gi
Node Affinity:   <none>
Message:        
Source:
    Type:       AWSElasticBlockStore (a Persistent Disk resource in AWS)
    VolumeID:   aws://us-east-1b/vol-xxxxx
    FSType:     ext4
    Partition:  0
    ReadOnly:   false
Events:         <none>

And the pod output:

$ kubectl describe pods

Name:               my-project-db-7d48567b48-slncd
Namespace:          my-project
Priority:           0
PriorityClassName:  <none>
Node:               ip-192-168-212-194.ec2.internal/192.168.212.194
Start Time:         Wed, 19 Dec 2018 15:55:58 +0100
Labels:             name=my-db
                    pod-template-hash=3804123604
Annotations:        <none>
Status:             Running
IP:                 192.168.216.33
Controlled By:      ReplicaSet/my-project-db-7d48567b48
Containers:
  my-project-db-container:
    Container ID:   docker://cf8222f15e395b02805c628b6addde2d77de2245aed9406a48c7c6f4dccefd4e
    Image:          mongo
    Image ID:       docker-pullable://mongo@sha256:0823cc2000223420f88b20d5e19e6bc252fa328c30d8261070e4645b02183c6a
    Port:           27017/TCP
    Host Port:      0/TCP
    State:          Running
      Started:      Wed, 19 Dec 2018 15:56:42 +0100
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /my_dir from my-project-db-claim0 (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-pf9ks (ro)
Conditions:
  Type           Status
  Initialized    True
  Ready          True
  PodScheduled   True
Volumes:
  my-project-db-claim0:
    Type:       PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
    ClaimName:  my-project-db-claim0
    ReadOnly:   false
  default-token-pf9ks:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-pf9ks
    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        7m22s (x5 over 7m23s)  default-scheduler                         pod has unbound PersistentVolumeClaims (repeated 2 times)
  Normal   Scheduled               7m21s                  default-scheduler                         Successfully assigned my-project/my-project-db-7d48567b48-slncd to ip-192-168-212-194.ec2.internal
  Normal   SuccessfulMountVolume   7m21s                  kubelet, ip-192-168-212-194.ec2.internal  MountVolume.SetUp succeeded for volume "default-token-pf9ks"
  Warning  FailedAttachVolume      7m13s (x5 over 7m21s)  attachdetach-controller                   AttachVolume.Attach failed for volume "pvc-307b755a-039e-11e9-b78d-0a68bcb24bc6" : "Error attaching EBS volume \"vol-01a863d0aa7c7e342\"" to instance "i-0a7dafbbdfeabc50b" since volume is in "creating" state
  Normal   SuccessfulAttachVolume  7m1s                   attachdetach-controller                   AttachVolume.Attach succeeded for volume "pvc-307b755a-039e-11e9-b78d-0a68bcb24bc6"
  Normal   SuccessfulMountVolume   6m48s                  kubelet, ip-192-168-212-194.ec2.internal  MountVolume.SetUp succeeded for volume "pvc-307b755a-039e-11e9-b78d-0a68bcb24bc6"
  Normal   Pulling                 6m48s                  kubelet, ip-192-168-212-194.ec2.internal  pulling image "mongo"
  Normal   Pulled                  6m39s                  kubelet, ip-192-168-212-194.ec2.internal  Successfully pulled image "mongo"
  Normal   Created                 6m38s                  kubelet, ip-192-168-212-194.ec2.internal  Created container
  Normal   Started                 6m37s                  kubelet, ip-192-168-212-194.ec2.internal  Started container

Everything is created without any problems. But if I uncomment the lines in the yaml so two volumes are attached to the db deployment, the pv output is the same as earlier, but the pod gets stuck on pending with the following output:

$ kubectl describe pods

Name:               my-project-db-b8b8d8bcb-l64d7
Namespace:          my-project
Priority:           0
PriorityClassName:  <none>
Node:               <none>
Labels:             name=my-db
                    pod-template-hash=646484676
Annotations:        <none>
Status:             Pending
IP:                 
Controlled By:      ReplicaSet/my-project-db-b8b8d8bcb
Containers:
  my-project-db-container:
    Image:        mongo
    Port:         27017/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:
      /data/db from my-project-db-claim1 (rw)
      /my_dir from my-project-db-claim0 (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-pf9ks (ro)
Conditions:
  Type           Status
  PodScheduled   False 
Volumes:
  my-project-db-claim0:
    Type:       PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
    ClaimName:  my-project-db-claim0
    ReadOnly:   false
  my-project-db-claim1:
    Type:       PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace)
    ClaimName:  my-project-db-claim1
    ReadOnly:   false
  default-token-pf9ks:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-pf9ks
    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  60s (x5 over 60s)  default-scheduler  pod has unbound PersistentVolumeClaims (repeated 2 times)
  Warning  FailedScheduling  2s (x16 over 59s)  default-scheduler  0/2 nodes are available: 2 node(s) had no available volume zone.

I've already read these two issues:

Dynamic volume provisioning creates EBS volume in the wrong availability zone

PersistentVolume on EBS can be created in availability zones with no nodes (Closed)

But I already checked that the volumes are created in the same zones as the cluster nodes instances. In fact, EKS creates two EBS by default in us-east-1b and us-east-1c zones and those volumes works. The volumes created by the posted yaml are on those regions too.


回答1:


Sounds like it's trying to create a volume in an availability zone where you don't have any volumes on. You can try restricting your StorageClass to the availability zones where you have nodes.

kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
  name: gp2
  annotations:
    storageclass.kubernetes.io/is-default-class: "true"
provisioner: kubernetes.io/aws-ebs
parameters:
  type: gp2
reclaimPolicy: Retain
mountOptions:
  - debug
allowedTopologies:
- matchLabelExpressions:
  - key: failure-domain.beta.kubernetes.io/zone
    values:
    - us-east-1b
    - us-east-1c

This is very similar to this question and this answer except that the issue described is on GCP and in this case it's AWS.




回答2:


See this article: https://kubernetes.io/blog/2018/10/11/topology-aware-volume-provisioning-in-kubernetes/

The gist is that you want to update your storageclass to include:

volumeBindingMode: WaitForFirstConsumer

This causes the PV to not be created until the pod is scheduled. It fixed a similar problem for me.




回答3:


In this case, you should check the availability zone of your worker nodes (EC2 instances).

As a Example :

worker node 1 = eu-central-1b
worker node 2 = eu-central-1c

Then create the volume in including one of an availability zone which mentioned above(do not create the volume with eu-central-1a).

after you create the volume, create your PersistentVolume and PersistentVolumeClaim by attaching a newly created volume to your cluster like below.

apiVersion: v1
kind: PersistentVolume
metadata:
  labels:
    failure-domain.beta.kubernetes.io/region: eu-central-1
    failure-domain.beta.kubernetes.io/zone: eu-central-1b
  name: mongo-pv
  namespace: default
spec:
  accessModes:
  - ReadWriteOnce
  capacity:
    storage: 100Gi
  awsElasticBlockStore:
    fsType: ext4
    volumeID: aws://eu-central-1b/vol-063342ab9be5d2929

  storageClassName: gp2

---

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: mongo-pvc
  namespace: default
spec:
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 100Gi
    storageClassName: gp2
    volumeName: mongo-pv


来源:https://stackoverflow.com/questions/53854588/kubernetes-pod-pending-when-a-new-volume-is-attached-eks

易学教程内所有资源均来自网络或用户发布的内容,如有违反法律规定的内容欢迎反馈
该文章没有解决你所遇到的问题?点击提问,说说你的问题,让更多的人一起探讨吧!