|
| 1 | +## Dynamic Provisioning |
| 2 | +**Important** |
| 3 | +You can't use dynamic provisioning with Fargate nodes. |
| 4 | + |
| 5 | +This example shows how to create a dynamically provisioned volume created through a directory on the file system and a Persistent Volume Claim (PVC) and consume it from a pod. |
| 6 | + |
| 7 | +**Prerequisite** |
| 8 | +This example requires Kubernetes 1.17 or later and a driver version of 1.5.x or later. |
| 9 | + |
| 10 | +1. Create a storage class for Amazon EFS. |
| 11 | + |
| 12 | + 1. Retrieve your Amazon EFS file system ID. You can find this in the Amazon EFS console, or use the following AWS CLI command. |
| 13 | + |
| 14 | + ```sh |
| 15 | + aws efs describe-file-systems --query "FileSystems[*].FileSystemId" --output text |
| 16 | + ``` |
| 17 | + |
| 18 | + The example output is as follows. |
| 19 | + |
| 20 | + ``` |
| 21 | + fs-582a03f3 |
| 22 | + ``` |
| 23 | + |
| 24 | + 2. Download a `StorageClass` manifest for Amazon EFS. |
| 25 | + |
| 26 | + ```sh |
| 27 | + curl -O https://raw.githubusercontent.com/kubernetes-sigs/aws-efs-csi-driver/master/examples/kubernetes/dynamic_provisioning/directories/specs/storageclass.yaml |
| 28 | + ``` |
| 29 | + |
| 30 | + 3. Edit [the file](./specs/storageclass.yaml). Find the following line, and replace the value for `fileSystemId` with your file system ID. |
| 31 | + |
| 32 | + ``` |
| 33 | + fileSystemId: fs-582a03f3 |
| 34 | + ``` |
| 35 | + |
| 36 | + Modify the other values as needed: |
| 37 | + * `fileSystemId` - The file system under which the access point is created. |
| 38 | + * `directoryPerms` - The directory permissions of the root directory created by the access point. |
| 39 | + * `basePath` (Optional) - The path on the file system under which the access point root directory is created. If the path isn't provided, the access points root directory is created under the root of the file system. |
| 40 | +
|
| 41 | + 4. Deploy the storage class. |
| 42 | +
|
| 43 | + ```sh |
| 44 | + kubectl apply -f storageclass.yaml |
| 45 | + ``` |
| 46 | +
|
| 47 | +2. Test automatic provisioning by deploying a Pod that makes use of the PVC: |
| 48 | +
|
| 49 | + 1. Download a manifest that deploys a Pod and a PVC. |
| 50 | +
|
| 51 | + ```sh |
| 52 | + curl -O https://raw.githubusercontent.com/kubernetes-sigs/aws-efs-csi-driver/master/examples/kubernetes/dynamic_provisioning/specs/pod.yaml |
| 53 | + ``` |
| 54 | +
|
| 55 | + 2. Deploy the Pod with a sample app and the PVC used by the Pod. |
| 56 | +
|
| 57 | + ```sh |
| 58 | + kubectl apply -f pod.yaml |
| 59 | + ``` |
| 60 | +
|
| 61 | +3. Determine the names of the Pods running the controller. |
| 62 | +
|
| 63 | + ```sh |
| 64 | + kubectl get pods -n kube-system | grep efs-csi-controller |
| 65 | + ``` |
| 66 | +
|
| 67 | + The example output is as follows. |
| 68 | +
|
| 69 | + ``` |
| 70 | + efs-csi-controller-74ccf9f566-q5989 3/3 Running 0 40m |
| 71 | + efs-csi-controller-74ccf9f566-wswg9 3/3 Running 0 40m |
| 72 | + ``` |
| 73 | +
|
| 74 | +4. After few seconds, you can observe the controller picking up the change \(edited for readability\). Replace `74ccf9f566-q5989` with a value from one of the Pods in your output from the previous command. |
| 75 | +
|
| 76 | + ```sh |
| 77 | + kubectl logs efs-csi-controller-74ccf9f566-q5989 \ |
| 78 | + -n kube-system \ |
| 79 | + -c csi-provisioner \ |
| 80 | + --tail 10 |
| 81 | + ``` |
| 82 | +
|
| 83 | + The example output is as follows. |
| 84 | +
|
| 85 | + ``` |
| 86 | + [...] |
| 87 | + 1 controller.go:737] successfully created PV pvc-5983ffec-96cf-40c1-9cd6-e5686ca84eca for PVC efs-claim... |
| 88 | + ``` |
| 89 | +
|
| 90 | + If you don't see the previous output, run the previous command using one of the other controller Pods. |
| 91 | + |
| 92 | +5. Confirm that a persistent volume was created with a status of `Bound` to a `PersistentVolumeClaim`: |
| 93 | + |
| 94 | + ```sh |
| 95 | + kubectl get pv |
| 96 | + ``` |
| 97 | + |
| 98 | + The example output is as follows. |
| 99 | + |
| 100 | + ``` |
| 101 | + NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE |
| 102 | + pvc-5983ffec-96cf-40c1-9cd6-e5686ca84eca 20Gi RWX Delete Bound default/efs-claim efs-sc 7m57s |
| 103 | + ``` |
| 104 | +
|
| 105 | +6. View details about the `PersistentVolumeClaim` that was created. |
| 106 | +
|
| 107 | + ```sh |
| 108 | + kubectl get pvc |
| 109 | + ``` |
| 110 | + |
| 111 | + The example output is as follows. |
| 112 | + |
| 113 | + ``` |
| 114 | + NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE |
| 115 | + efs-claim Bound pvc-5983ffec-96cf-40c1-9cd6-e5686ca84eca 20Gi RWX efs-sc 9m7s |
| 116 | + ``` |
| 117 | + |
| 118 | +7. View the sample app Pod's status until the `STATUS` becomes `Running`. |
| 119 | + |
| 120 | + ```sh |
| 121 | + kubectl get pods -o wide |
| 122 | + ``` |
| 123 | + |
| 124 | + The example output is as follows. |
| 125 | + |
| 126 | + ``` |
| 127 | + NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES |
| 128 | + efs-app 1/1 Running 0 10m 192.168.78.156 ip-192-168-73-191.region-code.compute.internal <none> <none> |
| 129 | + ``` |
| 130 | +**Note** |
| 131 | +If a Pod doesn't have an IP address listed, make sure that you added a mount target for the subnet that your node is in \(as described at the end of [Create an Amazon EFS file system](#efs-create-filesystem)\). Otherwise the Pod won't leave `ContainerCreating` status. When an IP address is listed, it may take a few minutes for a Pod to reach the `Running` status. |
| 132 | + |
| 133 | +1. Confirm that the data is written to the volume. |
| 134 | + |
| 135 | + ```sh |
| 136 | + kubectl exec efs-app -- bash -c "cat data/out" |
| 137 | + ``` |
| 138 | + |
| 139 | + The example output is as follows. |
| 140 | + |
| 141 | + ``` |
| 142 | + [...] |
| 143 | + Tue Mar 23 14:29:16 UTC 2021 |
| 144 | + Tue Mar 23 14:29:21 UTC 2021 |
| 145 | + Tue Mar 23 14:29:26 UTC 2021 |
| 146 | + Tue Mar 23 14:29:31 UTC 2021 |
| 147 | + [...] |
| 148 | + ``` |
| 149 | + |
| 150 | +2. \(Optional\) Terminate the Amazon EKS node that your Pod is running on and wait for the Pod to be re\-scheduled. Alternately, you can delete the Pod and redeploy it. Complete the previous step again, confirming that the output includes the previous output. |
0 commit comments