| copyright | lastupdated | keywords | subcollection | ||
|---|---|---|---|---|---|
|
2020-01-14 |
kubernetes, iks |
containers |
{:codeblock: .codeblock} {:deprecated: .deprecated} {:download: .download} {:external: target="_blank" .external} {:faq: data-hd-content-type='faq'} {:gif: data-image-type='gif'} {:help: data-hd-content-type='help'} {:important: .important} {:new_window: target="_blank"} {:note: .note} {:pre: .pre} {:preview: .preview} {:screen: .screen} {:shortdesc: .shortdesc} {:support: data-reuse='support'} {:table: .aria-labeledby="caption"} {:tip: .tip} {:troubleshoot: data-hd-content-type='troubleshoot'} {:tsCauses: .tsCauses} {:tsResolve: .tsResolve} {:tsSymptoms: .tsSymptoms}
{: #block_storage}
{{site.data.keyword.cloud_notm}} {{site.data.keyword.blockstorageshort}} is persistent, high-performance iSCSI storage that you can add to your apps by using Kubernetes persistent volumes (PVs). You can choose between predefined storage tiers with GB sizes and IOPS that meet the requirements of your workloads. To find out whether {{site.data.keyword.cloud_notm}} {{site.data.keyword.blockstorageshort}} is the right storage option for you, see Choosing a storage solution. For more information about pricing, see Billing. {: shortdesc}
{{site.data.keyword.cloud_notm}} {{site.data.keyword.blockstorageshort}} is available only for standard {{site.data.keyword.containerlong_notm}} clusters that are provisioned on classic infrastructure, and is not supported in VPC on Classic clusters. If your cluster cannot access the public network, such as a private cluster behind a firewall or a cluster with only the private service endpoint enabled, make sure that you installed the {{site.data.keyword.cloud_notm}} {{site.data.keyword.blockstorageshort}} plug-in version 1.3.0 or later to connect to your block storage instance over the private network. {{site.data.keyword.blockstorageshort}} instances are specific to a single zone. If you have a multizone cluster, consider multizone persistent storage options. {: important}
{: #install_block}
Install the {{site.data.keyword.cloud_notm}} Block Storage plug-in with a Helm chart to set up pre-defined storage classes for block storage. You can use these storage classes to create a PVC to provision block storage for your apps. {: shortdesc}
Before you begin: Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
-
Follow the instructions{: new_window} to install the Helm client on your local machine, and install the Helm server (Tiller) with a service account in your cluster.
The installation of the Helm server Tiller requires public network connection to the public Google Container Registry. If your cluster cannot access the public network, such as a private cluster behind a firewall or a cluster with only the private service endpoint that is enabled, you can choose to pull the Tiller image to your local machine, tag the image, and push it to your namespace in {{site.data.keyword.registryshort_notm}}{: new_window}. Or you can install the Helm chart without using Tiller{: new_window}. {: note}
-
Verify that Tiller is installed with a service account.
kubectl get serviceaccount -n kube-system tiller{: pre}
Example output:
NAME SECRETS AGE tiller 1 2m{: screen}
-
Add the {{site.data.keyword.cloud_notm}} Helm chart repository to the cluster where you want to use the {{site.data.keyword.cloud_notm}} Block Storage plug-in.
helm repo add iks-charts https://icr.io/helm/iks-charts{: pre}
-
Update the Helm repo to retrieve the latest version of all Helm charts in this repo.
helm repo update{: pre}
-
Install the {{site.data.keyword.cloud_notm}} Block Storage plug-in. When you install the plug-in, pre-defined block storage classes are added to your cluster.
helm install iks-charts/ibmcloud-block-storage-plugin{: pre}
Example output:
NAME: bald-olm LAST DEPLOYED: Wed Apr 18 10:02:55 2018 NAMESPACE: default STATUS: DEPLOYED RESOURCES: ==> v1beta1/DaemonSet NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE ibmcloud-block-storage-driver 0 0 0 0 0 <none> 0s ==> v1beta1/Deployment NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE ibmcloud-block-storage-plugin 1 0 0 0 0s ==> v1/StorageClass NAME PROVISIONER AGE ibmc-block-bronze ibm.io/ibmc-block 0s ibmc-block-custom ibm.io/ibmc-block 0s ibmc-block-gold ibm.io/ibmc-block 0s ibmc-block-retain-bronze ibm.io/ibmc-block 0s ibmc-block-retain-custom ibm.io/ibmc-block 0s ibmc-block-retain-gold ibm.io/ibmc-block 0s ibmc-block-retain-silver ibm.io/ibmc-block 0s ibmc-block-silver ibm.io/ibmc-block 0s ==> v1/ServiceAccount NAME SECRETS AGE ibmcloud-block-storage-plugin 1 0s ==> v1beta1/ClusterRole NAME AGE ibmcloud-block-storage-plugin 0s ==> v1beta1/ClusterRoleBinding NAME AGE ibmcloud-block-storage-plugin 0s NOTES: Thank you for installing: ibmcloud-block-storage-plugin. Your release is named: bald-olm{: screen}
-
Verify that the installation was successful.
kubectl get pod -n kube-system | grep block{: pre}
Example output:
ibmcloud-block-storage-driver-kh4mt 1/1 Running 0 27d 10.118.98.19 10.118.98.19 ibmcloud-block-storage-plugin-58c5f9dc86-pbl4t 1/1 Running 0 14d 172.21.0.204 10.118.98.19{: screen}
The installation is successful when you see one
ibmcloud-block-storage-pluginpod and one or moreibmcloud-block-storage-driverpods. The number ofibmcloud-block-storage-driverpods equals the number of worker nodes in your cluster. All pods must be in a Running state. -
Verify that the storage classes for block storage were added to your cluster.
kubectl get storageclasses | grep block{: pre}
Example output:
ibmc-block-bronze ibm.io/ibmc-block ibmc-block-custom ibm.io/ibmc-block ibmc-block-gold ibm.io/ibmc-block ibmc-block-retain-bronze ibm.io/ibmc-block ibmc-block-retain-custom ibm.io/ibmc-block ibmc-block-retain-gold ibm.io/ibmc-block ibmc-block-retain-silver ibm.io/ibmc-block ibmc-block-silver ibm.io/ibmc-block{: screen}
-
Repeat these steps for every cluster where you want to provision block storage.
You can now continue to create a PVC to provision block storage for your app.
You can upgrade the existing {{site.data.keyword.cloud_notm}} Block Storage plug-in to the latest version. {: shortdesc}
Before you begin: Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
-
Update the Helm repo to retrieve the latest version of all helm charts in this repo.
helm repo update{: pre}
-
Optional: Download the latest Helm chart to your local machine. Then, extract the package and review the
release.mdfile to find the latest release information.helm fetch iks-charts/ibmcloud-block-storage-plugin{: pre}
-
Find the name of the block storage Helm chart that you installed in your cluster.
helm ls | grep ibmcloud-block-storage-plugin{: pre}
Example output:
myhelmchart 1 Mon Sep 18 15:31:40 2017 DEPLOYED ibmcloud-block-storage-plugin-0.1.0 default{: screen}
-
Upgrade the {{site.data.keyword.cloud_notm}} Block Storage plug-in to the latest version.
helm upgrade --force --recreate-pods <helm_chart_name> iks-charts/ibmcloud-block-storage-plugin{: pre}
-
Optional: When you update the plug-in, the
defaultstorage class is unset. If you want to set the default storage class to a storage class of your choice, run the following command.kubectl patch storageclass <storageclass> -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'{: pre}
If you do not want to provision and use {{site.data.keyword.cloud_notm}} Block Storage in your cluster, you can uninstall the Helm chart. {: shortdesc}
Removing the plug-in does not remove existing PVCs, PVs, or data. When you remove the plug-in, all the related pods and daemon sets are removed from your cluster. You cannot provision new block storage for your cluster or use existing block storage PVCs and PVs after you remove the plug-in. {: important}
Before you begin:
- Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
- Make sure that you do not have any PVCs or PVs in your cluster that use block storage.
To remove the plug-in:
-
Find the name of the block storage Helm chart that you installed in your cluster.
helm ls | grep ibmcloud-block-storage-plugin{: pre}
Example output:
myhelmchart 1 Mon Sep 18 15:31:40 2017 DEPLOYED ibmcloud-block-storage-plugin-0.1.0 default{: screen}
-
Delete the {{site.data.keyword.cloud_notm}} Block Storage plug-in.
helm delete <helm_chart_name>{: pre}
-
Verify that the block storage pods are removed.
kubectl get pod -n kube-system | grep ibmcloud-block-storage-plugin{: pre} The removal of the pods is successful if no pods are displayed in your CLI output.
-
Verify that the block storage classes are removed.
kubectl get storageclasses | grep block{: pre} The removal of the storage classes is successful if no storage classes are displayed in your CLI output.
{: #block_predefined_storageclass}
{{site.data.keyword.containerlong_notm}} provides pre-defined storage classes for block storage that you can use to provision block storage with a specific configuration. {: shortdesc}
Every storage class specifies the type of block storage that you provision, including available size, IOPS, file system, and the retention policy.
Make sure to choose your storage configuration carefully to have enough capacity to store your data. After you provision a specific type of storage by using a storage class, you cannot change the type or retention policy for the storage device. However, you can change the size and the IOPS if you want to increase your storage capacity and performance. To change the type and retention policy for your storage, you must create a new storage instance and copy the data from the old storage instance to your new one. {: important}
-
List available storage classes in {{site.data.keyword.containerlong}}.
kubectl get storageclasses | grep block{: pre}
Example output:
$ kubectl get storageclasses NAME TYPE ibmc-block-custom ibm.io/ibmc-block ibmc-block-bronze ibm.io/ibmc-block ibmc-block-gold ibm.io/ibmc-block ibmc-block-silver ibm.io/ibmc-block ibmc-block-retain-bronze ibm.io/ibmc-block ibmc-block-retain-silver ibm.io/ibmc-block ibmc-block-retain-gold ibm.io/ibmc-block ibmc-block-retain-custom ibm.io/ibmc-block{: screen}
-
Review the configuration of a storage class.
kubectl describe storageclass <storageclass_name>{: pre}
For more information about each storage class, see the storage class reference. If you do not find what you are looking for, consider creating your own customized storage class. To get started, check out the customized storage class samples. {: tip}
-
Choose the type of block storage that you want to provision.
- Bronze, silver, and gold storage classes: These storage classes provision Endurance storage. With Endurance storage, you can choose the size of the storage in gigabytes at predefined IOPS tiers.
- Custom storage class: This storage class provisions Performance storage. With performance storage, you have more control over the size of the storage and the IOPS.
-
Choose the size and IOPS for your block storage. The size and the number of IOPS define the total number of IOPS (input/ output operations per second) that serves as an indicator for how fast your storage is. The more total IOPS your storage has, the faster it processes read and write operations.
- Bronze, silver, and gold storage classes: These storage classes come with a fixed number of IOPS per gigabyte and are provisioned on SSD hard disks. The total number of IOPS depends on the size of the storage that you choose. You can select any whole number of gigabyte within the allowed size range, such as 20 Gi, 256 Gi, or 11854 Gi. To determine the total number of IOPS, you must multiply the IOPS with the selected size. For example, if you select a 1000Gi block storage size in the silver storage class that comes with 4 IOPS per GB, your storage has a total of 4000 IOPS.
Table of storage class size ranges and IOPS per gigabyte Storage class IOPS per gigabyte Size range in gigabytes Bronze 2 IOPS/GB 20-12000 Gi Silver 4 IOPS/GB 20-12000 Gi Gold 10 IOPS/GB 20-4000 Gi - Custom storage class: When you choose this storage class, you have more control over the size and IOPS that you want. For the size, you can select any whole number of gigabyte within the allowed size range. The size that you choose determines the IOPS range that is available to you. You can choose an IOPS that is a multiple of 100 that is in the specified range. The IOPS that you choose is static and does not scale with the size of the storage. For example, if you choose 40Gi with 100 IOPS, your total IOPS remains 100.
The IOPS to gigabyte ratio also determines the type of hard disk that is provisioned for you. For example, if you have 500Gi at 100 IOPS, your IOPS to gigabyte ratio is 0.2. Storage with a ratio of less than or equal to 0.3 is provisioned on SATA hard disks. If your ratio is greater than 0.3, then your storage is provisioned on SSD hard disks.Table of custom storage class size ranges and IOPS Size range in gigabytes IOPS range in multiples of 100 20-39 Gi 100-1000 IOPS 40-79 Gi 100-2000 IOPS 80-99 Gi 100-4000 IOPS 100-499 Gi 100-6000 IOPS 500-999 Gi 100-10000 IOPS 1000-1999 Gi 100-20000 IOPS 2000-2999 Gi 200-40000 IOPS 3000-3999 Gi 200-48000 IOPS 4000-7999 Gi 300-48000 IOPS 8000-9999 Gi 500-48000 IOPS 10000-12000 Gi 1000-48000 IOPS
- Bronze, silver, and gold storage classes: These storage classes come with a fixed number of IOPS per gigabyte and are provisioned on SSD hard disks. The total number of IOPS depends on the size of the storage that you choose. You can select any whole number of gigabyte within the allowed size range, such as 20 Gi, 256 Gi, or 11854 Gi. To determine the total number of IOPS, you must multiply the IOPS with the selected size. For example, if you select a 1000Gi block storage size in the silver storage class that comes with 4 IOPS per GB, your storage has a total of 4000 IOPS.
-
Choose if you want to keep your data after the cluster or the persistent volume claim (PVC) is deleted.
- If you want to keep your data, then choose a
retainstorage class. When you delete the PVC, only the PVC is deleted. The PV, the physical storage device in your IBM Cloud infrastructure account, and your data still exist. To reclaim the storage and use it in your cluster again, you must remove the PV and follow the steps for using existing block storage. - If you want the PV, the data, and your physical block storage device to be deleted when you delete the PVC, choose a storage class without
retain.
- If you want to keep your data, then choose a
-
Choose if you want to be billed hourly or monthly. Check the pricing{: external} for more information. By default, all block storage devices are provisioned with an hourly billing type.
{: #add_block}
Create a persistent volume claim (PVC) to dynamically provision block storage for your cluster. Dynamic provisioning automatically creates the matching persistent volume (PV) and orders the actual storage device in your IBM Cloud infrastructure account. {:shortdesc}
Block storage comes with a ReadWriteOnce access mode. You can mount it to only one pod on one worker node in the cluster at a time.
{: note}
Before you begin:
- If you have a firewall, allow egress access for the IBM Cloud infrastructure IP ranges of the zones that your clusters are in so that you can create PVCs.
- Install the {{site.data.keyword.cloud_notm}} block storage plug-in.
- Decide on a pre-defined storage class or create a customized storage class.
Looking to deploy block storage in a stateful set? For more information, see Using block storage in a stateful set. {: tip}
To add block storage:
-
Create a configuration file to define your persistent volume claim (PVC) and save the configuration as a
.yamlfile.-
Example for bronze, silver, gold storage classes: The following
.yamlfile creates a claim that is namedmypvcof the"ibmc-block-silver"storage class, billed"hourly", with a gigabyte size of24Gi.apiVersion: v1 kind: PersistentVolumeClaim metadata: name: mypvc labels: billingType: "hourly" region: us-south zone: dal13 spec: accessModes: - ReadWriteOnce resources: requests: storage: 24Gi storageClassName: ibmc-block-silver
{: codeblock}
-
Example for using the custom storage class: The following
.yamlfile creates a claim that is namedmypvcof the storage classibmc-block-retain-custom, billed"hourly", with a gigabyte size of45Giand IOPS of"300".apiVersion: v1 kind: PersistentVolumeClaim metadata: name: mypvc labels: billingType: "hourly" region: us-south zone: dal13 spec: accessModes: - ReadWriteOnce resources: requests: storage: 45Gi iops: "300" storageClassName: ibmc-block-retain-custom
{: codeblock}
Understanding the YAML file components 
Understanding the YAML file componentsmetadata.nameEnter the name of the PVC. metadata.labels.billingTypeSpecify the frequency for which your storage bill is calculated, "monthly" or "hourly". The default is "hourly". metadata.labels.regionSpecify the region where you want to provision your block storage. If you specify the region, you must also specify a zone. If you do not specify a region, or the specified region is not found, the storage is created in the same region as your cluster. This option is supported only with the IBM Cloud Block Storage plug-in version 1.0.1 or higher. For older plug-in versions, if you have a multizone cluster, the zone in which your storage is provisioned is selected on a round-robin basis to balance volume requests evenly across all zones. To specify the zone for your storage, you can create a [customized storage class](#block_multizone_yaml) first. Then, create a PVC with your customized storage class.
metadata.labels.zoneSpecify the zone where you want to provision your block storage. If you specify the zone, you must also specify a region. If you do not specify a zone or the specified zone is not found in a multizone cluster, the zone is selected on a round-robin basis. This option is supported only with the IBM Cloud Block Storage plug-in version 1.0.1 or higher. For older plug-in versions, if you have a multizone cluster, the zone in which your storage is provisioned is selected on a round-robin basis to balance volume requests evenly across all zones. To specify the zone for your storage, you can create a [customized storage class](#block_multizone_yaml) first. Then, create a PVC with your customized storage class.
spec.resources.requests.storageEnter the size of the block storage, in gigabytes (Gi). After your storage is provisioned, you cannot change the size of your block storage. Make sure to specify a size that matches the amount of data that you want to store. spec.resources.requests.iopsThis option is available for the custom storage classes only (`ibmc-block-custom / ibmc-block-retain-custom`). Specify the total IOPS for the storage, selecting a multiple of 100 within the allowable range. If you choose an IOPS other than one that is listed, the IOPS is rounded up. spec.storageClassNameThe name of the storage class that you want to use to provision block storage. You can choose to use one of the [IBM-provided storage classes](#block_storageclass_reference) or [create your own storage class](#block_custom_storageclass).
If you do not specify a storage class, the PV is created with the default storage classibmc-file-bronze**Tip:** If you want to change the default storage class, run
kubectl patch storageclass <storageclass> -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/is-default-class":"true"}}}'and replace<storageclass>with the name of the storage class.If you want to use a customized storage class, create your PVC with the corresponding storage class name, a valid IOPS, and size.
{: tip} -
Create the PVC.
kubectl apply -f mypvc.yaml{: pre}
-
Verify that your PVC is created and bound to the PV. This process can take a few minutes.
kubectl describe pvc mypvc{: pre}
Example output:
Name: mypvc Namespace: default StorageClass: "" Status: Bound Volume: pvc-0d787071-3a67-11e7-aafc-eef80dd2dea2 Labels: <none> Capacity: 20Gi Access Modes: RWO Events: FirstSeen LastSeen Count From SubObjectPath Type Reason Message --------- -------- ----- ---- ------------- -------- ------ ------- 3m 3m 1 {ibm.io/ibmc-block 31898035-3011-11e7-a6a4-7a08779efd33 } Normal Provisioning External provisioner is provisioning volume for claim "default/my-persistent-volume-claim" 3m 1m 10 {persistentvolume-controller } Normal ExternalProvisioning cannot find provisioner "ibm.io/ibmc-block", expecting that a volume for the claim is provisioned either manually or via external software 1m 1m 1 {ibm.io/ibmc-block 31898035-3011-11e7-a6a4-7a08779efd33 } Normal ProvisioningSucceeded Successfully provisioned volume pvc-0d787071-3a67-11e7-aafc-eef80dd2dea2{: screen}
-
{: #block_app_volume_mount}To mount the PV to your deployment, create a configuration
.yamlfile and specify the PVC that binds the PV.apiVersion: apps/v1 kind: Deployment metadata: name: <deployment_name> labels: app: <deployment_label> spec: selector: matchLabels: app: <app_name> template: metadata: labels: app: <app_name> spec: containers: - image: <image_name> name: <container_name> volumeMounts: - name: <volume_name> mountPath: /<file_path> volumes: - name: <volume_name> persistentVolumeClaim: claimName: <pvc_name>
{: codeblock}
Understanding the YAML file components Understanding the YAML file componentsmetadata.labels.appA label for the deployment. spec.selector.matchLabels.app
spec.template.metadata.labels.appA label for your app. template.metadata.labels.appA label for the deployment. spec.containers.imageThe name of the image that you want to use. To list available images in your {{site.data.keyword.registryshort_notm}} account, run `ibmcloud cr image-list`. spec.containers.nameThe name of the container that you want to deploy to your cluster. spec.containers.volumeMounts.mountPathThe absolute path of the directory to where the volume is mounted inside the container. Data that is written to the mount path is stored under the root directory in your physical block storage instance. If you want to share a volume between different apps, you can specify [volume sub paths ](https://kubernetes.io/docs/concepts/storage/volumes/#using-subpath) for each of your apps. spec.containers.volumeMounts.nameThe name of the volume to mount to your pod. volumes.nameThe name of the volume to mount to your pod. Typically this name is the same as volumeMounts/name.volumes.persistentVolumeClaim.claimNameThe name of the PVC that binds the PV that you want to use. -
Create the deployment.
kubectl apply -f <local_yaml_path>{: pre}
-
Verify that the PV is successfully mounted.
kubectl describe deployment <deployment_name>{: pre}
The mount point is in the Volume Mounts field and the volume is in the Volumes field.
Volume Mounts: /var/run/secrets/kubernetes.io/serviceaccount from default-token-tqp61 (ro) /volumemount from myvol (rw) ... Volumes: myvol: Type: PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace) ClaimName: mypvc ReadOnly: false{: screen}
-
{: #existing_block}
If you have an existing physical storage device that you want to use in your cluster, you can manually create the PV and PVC to statically provision the storage. {: shortdesc}
Before you can start to mount your existing storage to an app, you must retrieve all necessary information for your PV.
{: #existing-block-1}
-
Retrieve or generate an API key for your IBM Cloud infrastructure account.
- Log in to the IBM Cloud infrastructure portal{: external}.
- Select Account, then Users, and then User List.
- Find your user ID.
- In the API KEY column, click Generate to generate an API key or View to view your existing API key.
-
Retrieve the API username for your IBM Cloud infrastructure account.
- From the User List menu, select your user ID.
- In the API Access Information section, find your API Username.
-
Log in to the IBM Cloud infrastructure CLI plug-in.
ibmcloud sl init{: pre}
-
Choose to authenticate by using the username and API key for your IBM Cloud infrastructure account.
-
Enter the username and API key that you retrieved in the previous steps.
-
List available block storage devices.
ibmcloud sl block volume-list{: pre}
Example output:
id username datacenter storage_type capacity_gb bytes_used ip_addr lunId active_transactions 38642141 IBM02SEL1543159-1 dal10 endurance_block_storage 20 - 169.xx.xxx.xxx 170 0{: screen}
-
Note the
id,ip_addr,capacity_gb, thedatacenter, andlunIdof the block storage device that you want to mount to your cluster. Note: To mount existing storage to a cluster, you must have a worker node in the same zone as your storage. To verify the zone of your worker node, runibmcloud ks worker ls --cluster <cluster_name_or_ID>.
{: #existing-block-2}
-
Optional: If you have storage that you provisioned with a
retainstorage class, when you remove the PVC, the PV and the physical storage device are not removed. To reuse the storage in your cluster, you must remove the PV first.-
List existing PVs.
kubectl get pv{: pre}
Look for the PV that belongs to your persistent storage. The PV is in a
releasedstate. -
Remove the PV.
kubectl delete pv <pv_name>{: pre}
-
Verify that the PV is removed.
kubectl get pv{: pre}
-
-
Create a configuration file for your PV. Include the block storage
id,ip_addr,capacity_gb, thedatacenter, andlunIdIDthat you retrieved earlier.apiVersion: v1 kind: PersistentVolume metadata: name: mypv labels: failure-domain.beta.kubernetes.io/region: <region> failure-domain.beta.kubernetes.io/zone: <zone> spec: capacity: storage: "<storage_size>" accessModes: - ReadWriteOnce flexVolume: driver: "ibm/ibmc-block" fsType: "<fs_type>" options: "Lun": "<lun_ID>" "TargetPortal": "<IP_address>" "VolumeID": "<volume_ID>" "volumeName": "<volume_name>"
{: codeblock}
Understanding the YAML file components Understanding the YAML file componentsmetadata.nameEnter the name of the PV that you want to create. metadata.labelsEnter the region and the zone that you retrieved earlier. You must have at least one worker node in the same region and zone as your persistent storage to mount the storage in your cluster. spec.flexVolume.fsTypeEnter the file system type that is configured for your existing block storage. Choose between ext4orxfs. If you do not specify this option, the PV defaults toext4. When the wrong `fsType` is defined, then the PV creation succeeds, but the mounting of the PV to a pod fails.spec.capacity.storageEnter the storage size of the existing block storage that you retrieved in the previous step as capacity-gb. The storage size must be written in gigabytes, for example, 20Gi (20 GB) or 1000Gi (1 TB).flexVolume.options.LunEnter the lun ID for your block storage that you retrieved earlier as lunId.flexVolume.options.TargetPortalEnter the IP address of your block storage that you retrieved earlier as ip_addr.flexVolume.options.VolumeIdEnter the ID of your block storage that you retrieved earlier as id.flexVolume.options.volumeNameEnter a name for your volume. -
Create the PV in your cluster.
kubectl apply -f mypv.yaml{: pre}
-
Verify that the PV is created.
kubectl get pv{: pre}
-
Create another configuration file to create your PVC. In order for the PVC to match the PV that you created earlier, you must choose the same value for
storageandaccessMode. Thestorage-classfield must be an empty string. If any of these fields do not match the PV, then a new PV is created automatically instead.kind: PersistentVolumeClaim apiVersion: v1 metadata: name: mypvc spec: accessModes: - ReadWriteOnce resources: requests: storage: "<storage_size>" storageClassName: ""
{: codeblock}
-
Create your PVC.
kubectl apply -f mypvc.yaml{: pre}
-
Verify that your PVC is created and bound to the PV that you created earlier. This process can take a few minutes.
kubectl describe pvc mypvc{: pre}
Example output:
Name: mypvc Namespace: default StorageClass: "" Status: Bound Volume: pvc-0d787071-3a67-11e7-aafc-eef80dd2dea2 Labels: <none> Capacity: 20Gi Access Modes: RWO Events: FirstSeen LastSeen Count From SubObjectPath Type Reason Message --------- -------- ----- ---- ------------- -------- ------ ------- 3m 3m 1 {ibm.io/ibmc-block 31898035-3011-11e7-a6a4-7a08779efd33 } Normal Provisioning External provisioner is provisioning volume for claim "default/my-persistent-volume-claim" 3m 1m 10 {persistentvolume-controller } Normal ExternalProvisioning cannot find provisioner "ibm.io/ibmc-block", expecting that a volume for the claim is provisioned either manually or via external software 1m 1m 1 {ibm.io/ibmc-block 31898035-3011-11e7-a6a4-7a08779efd33 } Normal ProvisioningSucceeded Successfully provisioned volume pvc-0d787071-3a67-11e7-aafc-eef80dd2dea2{: screen}
You successfully created a PV and bound it to a PVC. Cluster users can now mount the PVC to their deployments and start reading from and writing to the PV.
{: #block_statefulset}
If you have a stateful app such as a database, you can create stateful sets that use block storage to store your app's data. Alternatively, you can use an {{site.data.keyword.cloud_notm}} database-as-a-service and store your data in the cloud. {: shortdesc}
What do I need to be aware of when adding block storage to a stateful set?
To add storage to a stateful set, you specify your storage configuration in the volumeClaimTemplates section of your stateful set YAML. The volumeClaimTemplates is the basis for your PVC and can include the storage class and the size or IOPS of your block storage that you want to provision. However, if you want to include labels in your volumeClaimTemplates, Kubernetes does not include these labels when creating the PVC. Instead, you must add the labels directly to your stateful set.
You cannot deploy two stateful sets at the same time. If you try to create a stateful set before a different one is fully deployed, then the deployment of your stateful set might lead to unexpected results. {: important}
How can I create my stateful set in a specific zone?
In a multizone cluster, you can specify the zone and region where you want to create your stateful set in the spec.selector.matchLabels and spec.template.metadata.labels section of your stateful set YAML. Alternatively, you can add those labels to a customized storage class and use this storage class in the volumeClaimTemplates section of your stateful set.
Can I delay binding of a PV to my stateful pod until the pod is ready?
Yes, you can create a custom storage class for your PVC that includes the volumeBindingMode: WaitForFirstConsumer{: external} field.
What options do I have to add block storage to a stateful set?
If you want to automatically create your PVC when you create the stateful set, use dynamic provisioning. You can also choose to pre-provision your PVCs or use existing PVCs with your stateful set.
{: #block_dynamic_statefulset}
Use this option if you want to automatically create the PVC when you create the stateful set. {: shortdesc}
Before you begin: Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
-
Verify that all existing stateful sets in your cluster are fully deployed. If a stateful set is still being deployed, you cannot start creating your stateful set. You must wait until all stateful sets in your cluster are fully deployed to avoid unexpected results.
-
List existing stateful sets in your cluster.
kubectl get statefulset --all-namespaces{: pre}
Example output:
NAME DESIRED CURRENT AGE mystatefulset 3 3 6s{: screen}
-
View the Pods Status of each stateful set to ensure that the deployment of the stateful set is finished.
kubectl describe statefulset <statefulset_name>{: pre}
Example output:
Name: nginx Namespace: default CreationTimestamp: Fri, 05 Oct 2018 13:22:41 -0400 Selector: app=nginx,billingType=hourly,region=us-south,zone=dal10 Labels: app=nginx billingType=hourly region=us-south zone=dal10 Annotations: kubectl.kubernetes.io/last-applied-configuration={"apiVersion":"apps/v1","kind":"StatefulSet","metadata":{"annotations":{},"name":"nginx","namespace":"default"},"spec":{"podManagementPolicy":"Par... Replicas: 3 desired | 3 total Pods Status: 0 Running / 3 Waiting / 0 Succeeded / 0 Failed Pod Template: Labels: app=nginx billingType=hourly region=us-south zone=dal10 ...{: screen}
A stateful set is fully deployed when the number of replicas that you find in the Replicas section of your CLI output equals the number of Running pods in the Pods Status section. If a stateful set is not fully deployed yet, wait until the deployment is finished before you proceed.
-
-
Create a configuration file for your stateful set and the service that you use to expose the stateful set.
-
Example stateful set that specifies a zone:
The following example shows how to deploy NGINX as a stateful set with three replicas. For each replica, a 20 gigabyte block storage device is provisioned based on the specifications that are defined in the
ibmc-block-retain-bronzestorage class. All storage devices are provisioned in thedal10zone. Because block storage cannot be accessed from other zones, all replicas of the stateful set are also deployed onto worker nodes that are located indal10.apiVersion: v1 kind: Service metadata: name: nginx labels: app: nginx spec: ports: - port: 80 name: web clusterIP: None selector: app: nginx --- apiVersion: apps/v1 kind: StatefulSet metadata: name: nginx spec: serviceName: "nginx" replicas: 3 podManagementPolicy: Parallel selector: matchLabels: app: nginx billingType: "hourly" region: "us-south" zone: "dal10" template: metadata: labels: app: nginx billingType: "hourly" region: "us-south" zone: "dal10" spec: containers: - name: nginx image: k8s.gcr.io/nginx-slim:0.8 ports: - containerPort: 80 name: web volumeMounts: - name: myvol mountPath: /usr/share/nginx/html volumeClaimTemplates: - metadata: name: myvol spec: accessModes: - ReadWriteOnce resources: requests: storage: 20Gi iops: "300" #required only for performance storage storageClassName: ibmc-block-retain-bronze
{: codeblock}
-
Example stateful set with anti-affinity rule and delayed block storage creation:
The following example shows how to deploy NGINX as a stateful set with three replicas. The stateful set does not specify the region and zone where the block storage is created. Instead, the stateful set uses an anti-affinity rule to ensure that the pods are spread across worker nodes and zones. By defining
topologykey: failure-domain.beta.kubernetes.io/zone, the Kubernetes scheduler cannot schedule a pod on a worker node if the worker node is in the same zone as a pod that has theapp: nginxlabel. For each stateful set pod, two PVCs are created as defined in thevolumeClaimTemplatessection, but the creation of the block storage instances is delayed until a stateful set pod that uses the storage is scheduled. This setup is referred to as topology-aware volume scheduling.apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-bronze-delayed parameters: billingType: hourly classVersion: "2" fsType: ext4 iopsPerGB: "2" sizeRange: '[20-12000]Gi' type: Endurance provisioner: ibm.io/ibmc-block reclaimPolicy: Delete volumeBindingMode: WaitForFirstConsumer --- apiVersion: v1 kind: Service metadata: name: nginx labels: app: nginx spec: ports: - port: 80 name: web clusterIP: None selector: app: nginx --- apiVersion: apps/v1 kind: StatefulSet metadata: name: web spec: serviceName: "nginx" replicas: 3 podManagementPolicy: "Parallel" selector: matchLabels: app: nginx template: metadata: labels: app: nginx spec: affinity: podAntiAffinity: preferredDuringSchedulingIgnoredDuringExecution: - weight: 100 podAffinityTerm: labelSelector: matchExpressions: - key: app operator: In values: - nginx topologyKey: failure-domain.beta.kubernetes.io/zone containers: - name: nginx image: k8s.gcr.io/nginx-slim:0.8 ports: - containerPort: 80 name: web volumeMounts: - name: myvol1 mountPath: /usr/share/nginx/html - name: myvol2 mountPath: /tmp1 volumeClaimTemplates: - metadata: name: myvol1 spec: accessModes: - ReadWriteOnce # access mode resources: requests: storage: 20Gi storageClassName: ibmc-block-bronze-delayed - metadata: name: myvol2 spec: accessModes: - ReadWriteOnce # access mode resources: requests: storage: 20Gi storageClassName: ibmc-block-bronze-delayed
{: codeblock}
Understanding the stateful set YAML file components Understanding the stateful set YAML file componentsmetadata.nameEnter a name for your stateful set. The name that you enter is used to create the name for your PVC in the format: <volume_name>-<statefulset_name>-<replica_number>.spec.serviceNameEnter the name of the service that you want to use to expose your stateful set. spec.replicasEnter the number of replicas for your stateful set. spec.podManagementPolicyEnter the pod management policy that you want to use for your stateful set. Choose between the following options: - `OrderedReady`: With this option, stateful set replicas are deployed one after another. For example, if you specified three replicas, then Kubernetes creates the PVC for your first replica, waits until the PVC is bound, deploys the stateful set replica, and mounts the PVC to the replica. After the deployment is finished, the second replica is deployed. For more information about this option, see [`OrderedReady` Pod Management ](https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/#orderedready-pod-management).
- Parallel: With this option, the deployment of all stateful set replicas is started at the same time. If your app supports parallel deployment of replicas, then use this option to save deployment time for your PVCs and stateful set replicas.
spec.selector.matchLabelsEnter all labels that you want to include in your stateful set and your PVC. Labels that you include in the volumeClaimTemplatesof your stateful set are not recognized by Kubernetes. Sample labels that you might want to include are:regionandzone: If you want all your stateful set replicas and PVCs to be created in one specific zone, add both labels. You can also specify the zone and region in the storage class that you use. If you do not specify a zone and region and you have a multizone cluster, the zone in which your storage is provisioned is selected on a round-robin basis to balance volume requests evenly across all zones.billingType: Enter the billing type that you want to use for your PVCs. Choose betweenhourlyormonthly. If you do not specify this label, all PVCs are created with an hourly billing type.
spec.template.metadata.labelsEnter the same labels that you added to the spec.selector.matchLabelssection.spec.template.spec.affinitySpecify your anti-affinity rule to ensure that your stateful set pods are distributed across worker nodes and zones. The example shows an anti-affinity rule where the stateful set pod prefers not to be scheduled on a worker node where a pod runs that has the `app: nginx` label. The `topologykey: failure-domain.beta.kubernetes.io/zone` restricts this anti-affinity rule even more and prevents the pod to be scheduled on a worker node if the worker node is in the same zone as a pod that has the `app: nginx` label. By using this anti-affinity rule, you can achieve anti-affinity across worker nodes and zones. spec.volumeClaimTemplates.metadata.nameEnter a name for your volume. Use the same name that you defined in the spec.containers.volumeMount.namesection. The name that you enter here is used to create the name for your PVC in the format:<volume_name>-<statefulset_name>-<replica_number>.spec.volumeClaimTemplates.spec.resources.requests.storageEnter the size of the block storage in gigabytes (Gi). spec.volumeClaimTemplates.spec.resources.requests.iopsIf you want to provision [performance storage](#block_predefined_storageclass), enter the number of IOPS. If you use an endurance storage class and specify a number of IOPS, the number of IOPS is ignored. Instead, the IOPS that is specified in your storage class is used. spec.volumeClaimTemplates.spec.storageClassNameEnter the storage class that you want to use. To list existing storage classes, run kubectl get storageclasses | grep block. If you do not specify a storage class, the PVC is created with the default storage class that is set in your cluster. Make sure that the default storage class uses theibm.io/ibmc-blockprovisioner so that your stateful set is provisioned with block storage.
-
-
Create your stateful set.
kubectl apply -f statefulset.yaml{: pre}
-
Wait for your stateful set to be deployed.
kubectl describe statefulset <statefulset_name>{: pre}
To see the current status of your PVCs, run
kubectl get pvc. The name of your PVC is formatted as<volume_name>-<statefulset_name>-<replica_number>. {: tip}
{: #block_static_statefulset}
You can pre-provision your PVCs before creating your stateful set or use existing PVCs with your stateful set. {: shortdesc}
When you dynamically provision your PVCs when creating the stateful set, the name of the PVC is assigned based on the values that you used in the stateful set YAML file. In order for the stateful set to use existing PVCs, the name of your PVCs must match the name that would automatically be created when using dynamic provisioning.
Before you begin: Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
-
If you want to pre-provision the PVC for your stateful set before you create the stateful set, follow steps 1-3 in Adding block storage to apps to create a PVC for each stateful set replica. Make sure that you create your PVC with a name that follows the following format:
<volume_name>-<statefulset_name>-<replica_number>.<volume_name>: Use the name that you want to specify in thespec.volumeClaimTemplates.metadata.namesection of your stateful set, such asnginxvol.<statefulset_name>: Use the name that you want to specify in themetadata.namesection of your stateful set, such asnginx_statefulset.<replica_number>: Enter the number of your replica, starting with 0.
For example, if you must create three stateful set replicas, create three PVCs with the following names:
nginxvol-nginx_statefulset-0,nginxvol-nginx_statefulset-1, andnginxvol-nginx_statefulset-2.Looking to create a PVC and PV for an existing storage device? Create your PVC and PV by using static provisioning.
-
Follow the steps in Dynamic provisioning: Creating the PVC when you create a stateful set to create your stateful set. The name of your PVC follows the format
<volume_name>-<statefulset_name>-<replica_number>. Make sure to use the following values from your PVC name in the stateful set specification:spec.volumeClaimTemplates.metadata.name: Enter the<volume_name>of your PVC name.metadata.name: Enter the<statefulset_name>of your PVC name.spec.replicas: Enter the number of replicas that you want to create for your stateful set. The number of replicas must equal the number of PVCs that you created earlier.
If your PVCs are in different zones, do not include a region or zone label in your stateful set. {: note}
-
Verify that the PVCs are used in your stateful set replica pods.
-
List the pods in your cluster. Identify the pods that belong to your stateful set.
kubectl get pods{: pre}
-
Verify that your existing PVC is mounted to your stateful set replica. Review the
ClaimNamein theVolumessection of your CLI output.kubectl describe pod <pod_name>{: pre}
Example output:
Name: nginx-0 Namespace: default Node: 10.xxx.xx.xxx/10.xxx.xx.xxx Start Time: Fri, 05 Oct 2018 13:24:59 -0400 ... Volumes: myvol: Type: PersistentVolumeClaim (a reference to a PersistentVolumeClaim in the same namespace) ClaimName: myvol-nginx-0
...
{: screen} -
{: #block_change_storage_configuration}
If you want to increase storage capacity or performance, you can modify your existing volume. {: shortdesc}
For questions about billing and to find the steps for how to use the {{site.data.keyword.cloud_notm}} console to modify your storage, see Expanding Block Storage capacity and Adjusting IOPS. Updates that you make from the console are not reflected in the persistent volume (PV). To add this information to the PV, run kubectl patch pv <pv_name> and manually update the size and IOPS in the Labels and Annotation section of your PV.
{: tip}
-
List the PVCs in your cluster and note the name of the associated PV from the VOLUME column.
kubectl get pvc{: pre}
Example output:
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE myvol Bound pvc-01ac123a-123b-12c3-abcd-0a1234cb12d3 20Gi RWO ibmc-block-bronze 147d{: screen}
-
If you want to change the IOPS and the size for your block storage, edit the IOPS in the
metadata.labels.IOPSsection of your PV first. You can change to a lower or greater IOPS value. Make sure that you enter an IOPS that is supported for the storage type that you have. For example, if you have endurance block storage with 4 IOPS, you can change the IOPS to either 2 or 10. For more supported IOPS values, see Deciding on your block storage configuration.kubectl edit pv <pv_name>{: pre}
To change the IOPS from the CLI, you must also change the size of your block storage. If you want to change only the IOPS, but not the size, you must request the IOPS change from the console. {: note}
-
Edit the PVC and add the new size in the
spec.resources.requests.storagesection of your PVC. You can change to a greater size only up to the maximum capacity that is set by your storage class. You cannot downsize your existing storage. To see available sizes for your storage class, see Deciding on the block storage configuration.kubectl edit pvc <pvc_name>{: pre}
-
Verify that the volume expansion is requested. The volume expansion is successfully requested when you see a
FileSystemResizePendingmessage in the Conditions section of your CLI output.kubectl describe pvc <pvc_name>{: pre}
Example output:
... Conditions: Type Status LastProbeTime LastTransitionTime Reason Message ---- ------ ----------------- ------------------ ------ ------- FileSystemResizePending True Mon, 01 Jan 0001 00:00:00 +0000 Thu, 25 Apr 2019 15:52:49 -0400 Waiting for user to (re-)start a pod to finish file system resize of volume on node.{: screen}
-
List all the pods that mount the PVC. If your PVC is mounted by a pod, the volume expansion is automatically processed. If your PVC is not mounted by a pod, you must mount the PVC to a pod so that the volume expansion can be processed.
kubectl get pods --all-namespaces -o=jsonpath='{range .items[*]}{"\n"}{.metadata.name}{":\t"}{range .spec.volumes[*]}{.persistentVolumeClaim.claimName}{" "}{end}{end}' | grep "<pvc_name>"{: pre}
Mounted pods are returned in the format:
<pod_name>: <pvc_name>. -
If your PVC is not mounted by a pod, create a pod or deployment and mount the PVC. If your PVC is mounted by a pod, continue with the next step.
-
Verify that the size and IOPS are changed in the Labels section of your CLI output. This process might take a few minutes to complete.
kubectl describe pv <pv_name>{: pre}
Example output:
... Labels: CapacityGb=50 Datacenter=dal10 IOPS=500{: screen}
{: #block_backup_restore}
Block storage is provisioned into the same location as the worker nodes in your cluster. The storage is hosted on clustered servers by IBM to provide availability in case a server goes down. However, block storage is not backed up automatically and might be inaccessible if the entire location fails. To protect your data from being lost or damaged, you can set up periodic backups that you can use to restore your data when needed. {: shortdesc}
Review the following backup and restore options for your block storage:
- Set up periodic snapshots
You can [set up periodic snapshots for your block storage](/docs/infrastructure/BlockStorage?topic=BlockStorage-snapshots#snapshots), which is a read-only image that captures the state of the instance at a point in time. To store the snapshot, you must request snapshot space on your block storage. Snapshots are stored on the existing storage instance within the same zone. You can restore data from a snapshot if a user accidentally removes important data from the volume.
To create a snapshot for your volume:- [Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.](/docs/containers?topic=containers-cs_cli_install#cs_cli_configure)
- Log in to the `ibmcloud sl` CLI.
ibmcloud sl init - List existing PVs in your cluster.
kubectl get pv - Get the details for the PV for which you want to create snapshot space and note the volume ID, the size, and the IOPS. The size and IOPS are shown in the Labels section of your CLI output. To find the volume ID, review the
kubectl describe pv <pv_name>ibm.io/network-storage-idannotation of your CLI output. - Create the snapshot size for your existing volume with the parameters that you retrieved in the previous step.
ibmcloud sl block snapshot-order <volume_ID> --size <size> --tier <iops> - Wait for the snapshot size to create. The snapshot size is successfully provisioned when the Snapshot Size (GB) in your CLI output changes from 0 to the size that you ordered.
ibmcloud sl block volume-detail <volume_ID> - Create the snapshot for your volume and note the ID of the snapshot that is created for you.
ibmcloud sl block snapshot-create <volume_ID> - Verify that the snapshot is created successfully.
ibmcloud sl block snapshot-list <volume_ID>
To restore data from a snapshot to an existing volume:ibmcloud sl block snapshot-restore <volume_ID> <snapshot_ID>- Replicate snapshots to another zone
To protect your data from a zone failure, you can [replicate snapshots](/docs/infrastructure/BlockStorage?topic=BlockStorage-replication#replication) to a block storage instance that is set up in another zone. Data can be replicated from the primary storage to the backup storage only. You cannot mount a replicated block storage instance to a cluster. When your primary storage fails, you can manually set your replicated backup storage to be the primary one. Then, you can mount it to your cluster. After your primary storage is restored, you can restore the data from the backup storage.
- Duplicate storage
You can [duplicate your block storage instance](/docs/infrastructure/BlockStorage?topic=BlockStorage-duplicatevolume#duplicatevolume) in the same zone as the original storage instance. A duplicate has the same data as the original storage instance at the point in time that you create the duplicate. Unlike replicas, use the duplicate as an independent storage instance from the original. To duplicate, first set up snapshots for the volume.
- Back up data to {{site.data.keyword.cos_full}}
You can use the [**ibm-backup-restore image**](/docs/services/RegistryImages/ibm-backup-restore?topic=RegistryImages-ibmbackup_restore_starter#ibmbackup_restore_starter) to spin up a backup and restore pod in your cluster. This pod contains a script to run a one-time or periodic backup for any persistent volume claim (PVC) in your cluster. Data is stored in your {{site.data.keyword.cos_full}} instance that you set up in a zone.
Block storage is mounted with an RWO access mode. This access allows only one pod to be mounted to the block storage at a time. To back up your data, you must unmount the app pod from the storage, mount it to your backup pod, back up the data, and remount the storage to your app pod.
To make your data even more highly available and protect your app from a zone failure, set up a second {{site.data.keyword.cos_short}} instance and replicate data across zones. If you need to restore data from your {{site.data.keyword.cos_short}} instance, use the restore script that is provided with the image.- Copy data to and from pods and containers
You can use the `kubectl cp` [command](https://kubernetes.io/docs/reference/kubectl/overview/#cp) to copy files and directories to and from pods or specific containers in your cluster.
Before you begin: [Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.](/docs/containers?topic=containers-cs_cli_install#cs_cli_configure) If you do not specify a container with
-c, the command uses to the first available container in the pod.You can use the command in various ways:
- Copy data from your local machine to a pod in your cluster:
kubectl cp <local_filepath>/<filename> <namespace>/<pod>:<pod_filepath> - Copy data from a pod in your cluster to your local machine:
kubectl cp <namespace>/<pod>:<pod_filepath>/<filename> <local_filepath>/<filename> - Copy data from your local machine to a specific container that runs in a pod in your cluster:
kubectl cp <local_filepath>/<filename> <namespace>/<pod>:<pod_filepath> -c <container>
- Copy data from your local machine to a pod in your cluster:
{: #block_storageclass_reference}
| Characteristics | Setting |
|---|---|
| Name | ibmc-block-bronzeibmc-block-retain-bronze |
| Type | Endurance storage |
| File system | ext4 |
| IOPS per gigabyte | 2 |
| Size range in gigabytes | 20-12000 Gi |
| Hard disk | SSD |
| Billing | The default billing type depends on the version of your {{site.data.keyword.cloud_notm}} Block Storage plug-in:
|
| Pricing | Pricing information |
| {: class="simple-tab-table"} | |
| {: caption="Block storage class: bronze" caption-side="top"} | |
| {: #block_bronze} | |
| {: tab-title="Bronze"} | |
| {: tab-group="Block storage class"} |
| Characteristics | Setting |
|---|---|
| Name | ibmc-block-silveribmc-block-retain-silver |
| Type | Endurance storage |
| File system | ext4 |
| IOPS per gigabyte | 4 |
| Size range in gigabytes | 20-12000 Gi |
| Hard disk | SSD |
| Billing | The default billing type depends on the version of your {{site.data.keyword.cloud_notm}} Block Storage plug-in:
|
| Pricing | Pricing information |
| {: class="simple-tab-table"} | |
| {: caption="Block storage class: silver" caption-side="top"} | |
| {: #block_silver} | |
| {: tab-title="Silver"} | |
| {: tab-group="Block storage class"} |
| Characteristics | Setting |
|---|---|
| Name | ibmc-block-goldibmc-block-retain-gold |
| Type | Endurance storage |
| File system | ext4 |
| IOPS per gigabyte | 10 |
| Size range in gigabytes | 20-4000 Gi |
| Hard disk | SSD |
| Billing | The default billing type depends on the version of your {{site.data.keyword.cloud_notm}} Block Storage plug-in:
|
| Pricing | Pricing information |
| {: class="simple-tab-table"} | |
| {: caption="Block storage class: gold" caption-side="top"} | |
| {: #block_gold} | |
| {: tab-title="Gold"} | |
| {: tab-group="Block storage class"} |
| Characteristics | Setting |
|---|---|
| Name | ibmc-block-customibmc-block-retain-custom |
| Type | Performance |
| File system | ext4 |
| IOPS and size | Size range in gigabytes / IOPS range in multiples of 100
|
| Hard disk | The IOPS to gigabyte ratio determines the type of hard disk that is provisioned. To determine your IOPS to gigabyte ratio, you divide the IOPS by the size of your storage. Example: You chose 500Gi of storage with 100 IOPS. Your ratio is 0.2 (100 IOPS/500Gi). Overview of hard disk types per ratio:
|
| Billing | The default billing type depends on the version of your {{site.data.keyword.cloud_notm}} Block Storage plug-in:
|
| Pricing | Pricing information |
| {: class="simple-tab-table"} | |
| {: caption="Block storage class: custom" caption-side="top"} | |
| {: #block_custom} | |
| {: tab-title="Custom"} | |
| {: tab-group="Block storage class"} |
{: #block_custom_storageclass}
You can create a customized storage class and use the storage class in your PVC. {: shortdesc}
{{site.data.keyword.containerlong_notm}} provides pre-defined storage classes to provision block storage with a particular tier and configuration. In some cases, you might want to provision storage with a different configuration that is not covered in the pre-defined storage classes. You can use the examples in this topic to find sample customized storage classes.
To create your customized storage class, see Customizing a storage class. Then, use your customized storage class in your PVC.
{: #topology_yaml}
To use block storage in a multizone cluster, your pod must be scheduled in the same zone as your block storage instance so that you can read and write to the volume. Before topology-aware volume scheduling was introduced by Kubernetes, the dynamic provisioning of your storage automatically created the block storage instance when a PVC was created. Then, when you created your pod, the Kubernetes scheduler tried to deploy the pod to the same data center as your block storage instance. {: shortdesc}
Creating the block storage instance without knowing the constraints of the pod can lead to unwanted results. For example, your pod might not be able to be scheduled to the same worker node as your storage because the worker node has insufficient resources or the worker node is tainted and does not allow the pod to be scheduled. With topology-aware volume scheduling, the block storage instance is delayed until the first pod that uses the storage is created.
Topology-aware volume scheduling is supported on clusters that run Kubernetes version 1.12 or later only. To use this feature, make sure that you installed the {{site.data.keyword.cloud_notm}} Block Storage plug-in version 1.2.0 or later. {: note}
The following examples show how to create storage classes that delay the creation of the block storage instance until the first pod that uses this storage is ready to be scheduled. To delay the creation, you must include the volumeBindingMode: WaitForFirstConsumer option. If you do not include this option, the volumeBindingMode is automatically set to Immediate and the block storage instance is created when you create the PVC.
-
Example for Endurance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-bronze-delayed parameters: billingType: hourly classVersion: "2" fsType: ext4 iopsPerGB: "2" sizeRange: '[20-12000]Gi' type: Endurance provisioner: ibm.io/ibmc-block reclaimPolicy: Delete volumeBindingMode: WaitForFirstConsumer
{: codeblock}
-
Example for Performance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-performance-storageclass labels: kubernetes.io/cluster-service: "true" provisioner: ibm.io/ibmc-block parameters: billingType: "hourly" classVersion: "2" sizeIOPSRange: |- "[20-39]Gi:[100-1000]" "[40-79]Gi:[100-2000]" "[80-99]Gi:[100-4000]" "[100-499]Gi:[100-6000]" "[500-999]Gi:[100-10000]" "[1000-1999]Gi:[100-20000]" "[2000-2999]Gi:[200-40000]" "[3000-3999]Gi:[200-48000]" "[4000-7999]Gi:[300-48000]" "[8000-9999]Gi:[500-48000]" "[10000-12000]Gi:[1000-48000]" type: "Performance" reclaimPolicy: Delete volumeBindingMode: WaitForFirstConsumer
{: codeblock}
{: #block_multizone_yaml}
If you want to create your block storage in a specific zone, you can specify the zone and region in a customized storage class. {: shortdesc}
Use the customized storage class if you use the {{site.data.keyword.cloud_notm}} Block Storage plug-in version 1.0.0 or if you want to statically provision block storage in a specific zone. In all other cases, specify the zone directly in your PVC. {: note}
The following .yaml file customizes a storage class that is based on the ibm-block-silver non-retaining storage class: the type is "Endurance", the iopsPerGB is 4, the sizeRange is "[20-12000]Gi", and the reclaimPolicy is set to "Delete". The zone is specified as dal12. To use a different storage class as your base, see the storage class reference.
Create the storage class in the same region and zone as your cluster and worker nodes. To get the region of your cluster, run ibmcloud ks cluster get --cluster <cluster_name_or_ID> and look for the region prefix in the Master URL, such as eu-de in https://c2.eu-de.containers.cloud.ibm.com:11111. To get the zone of your worker node, run ibmcloud ks worker ls --cluster <cluster_name_or_ID>.
-
Example for Endurance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-silver-mycustom-storageclass labels: kubernetes.io/cluster-service: "true" provisioner: ibm.io/ibmc-block parameters: zone: "dal12" region: "us-south" type: "Endurance" iopsPerGB: "4" sizeRange: "[20-12000]Gi" reclaimPolicy: "Delete"
{: codeblock}
-
Example for Performance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-performance-storageclass labels: kubernetes.io/cluster-service: "true" provisioner: ibm.io/ibmc-block parameters: zone: "dal12" region: "us-south" type: "Performance" sizeIOPSRange: |- "[20-39]Gi:[100-1000]" "[40-79]Gi:[100-2000]" "[80-99]Gi:[100-4000]" "[100-499]Gi:[100-6000]" "[500-999]Gi:[100-10000]" "[1000-1999]Gi:[100-20000]" "[2000-2999]Gi:[200-40000]" "[3000-3999]Gi:[200-48000]" "[4000-7999]Gi:[300-48000]" "[8000-9999]Gi:[500-48000]" "[10000-12000]Gi:[1000-48000]" reclaimPolicy: "Delete"
{: codeblock}
{: #xfs}
The following examples create a storage class that provisions block storage with an XFS file system.
{: shortdesc}
-
Example for Endurance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-custom-xfs labels: addonmanager.kubernetes.io/mode: Reconcile provisioner: ibm.io/ibmc-block parameters: type: "Endurance" iopsPerGB: "4" sizeRange: "[20-12000]Gi" fsType: "xfs" reclaimPolicy: "Delete" -
Example for Performance block storage:
apiVersion: storage.k8s.io/v1 kind: StorageClass metadata: name: ibmc-block-custom-xfs labels: addonmanager.kubernetes.io/mode: Reconcile provisioner: ibm.io/ibmc-block parameters: type: "Performance" sizeIOPSRange: |- [20-39]Gi:[100-1000] [40-79]Gi:[100-2000] [80-99]Gi:[100-4000] [100-499]Gi:[100-6000] [500-999]Gi:[100-10000] [1000-1999]Gi:[100-20000] [2000-2999]Gi:[200-40000] [3000-3999]Gi:[200-48000] [4000-7999]Gi:[300-48000] [8000-9999]Gi:[500-48000] [10000-12000]Gi:[1000-48000] fsType: "xfs" reclaimPolicy: "Delete" classVersion: "2"
{: codeblock}
{: #cleanup}
When you set up persistent storage in your cluster, you have three main components: the Kubernetes persistent volume claim (PVC) that requests storage, the Kubernetes persistent volume (PV) that is mounted to a pod and described in the PVC, and the IBM Cloud infrastructure instance, such as classic file or block storage. Depending on how you created your storage, you might need to delete all three components separately. {:shortdesc}
{: #storage_delete_options}
Removing persistent storage from your {{site.data.keyword.cloud_notm}} account varies depending on how you provisioned the storage and what components you already removed. {: shortdesc}
Is my persistent storage deleted when I delete my cluster?
During cluster deletion, you have the option to remove your persistent storage. However, depending on how your storage was provisioned, the removal of your storage might not include all storage components.
If you dynamically provisioned storage with a storage class that sets reclaimPolicy: Delete, your PVC, PV, and the storage instance are automatically deleted when you delete the cluster. For storage that was statically provisioned, VPC Block Storage, or storage that you provisioned with a storage class that sets reclaimPolicy: Retain, the PVC and the PV are removed when you delete the cluster, but your storage instance and your data remain. You are still charged for your storage instance. Also, if you deleted your cluster in an unhealthy state, the storage might still exist even if you chose to remove it.
How do I delete the storage when I want to keep my cluster?
When you dynamically provisioned the storage with a storage class that sets reclaimPolicy: Delete, you can remove the PVC to start the deletion process of your persistent storage. Your PVC, PV, and storage instance are automatically removed.
For storage that was statically provisioned, VPC Block Storage, or storage that you provisioned with a storage class that sets reclaimPolicy: Retain, you must manually remove the PVC, PV, and the storage instance to avoid further charges.
How does the billing stop after I delete my storage?
Depending on what storage components you delete and when, the billing cycle might not stop immediately. If you delete the PVC and PV, but not the storage instance in your {{site.data.keyword.cloud_notm}} account, that instance still exists and you are charged for it.
If you delete the PVC, PV, and the storage instance, the billing cycle stops depending on the billingType that you chose when you provisioned your storage and how you chose to delete the storage.
-
When you manually cancel the persistent storage instance from the {{site.data.keyword.cloud_notm}} console or the
ibmcloud slCLI, billing stops as follows:- Hourly storage: Billing stops immediately. After your storage is canceled, you might still see your storage instance in the console for up to 72 hours.
- Monthly storage: You can choose between immediate cancellation or cancellation on the anniversary date. In both cases, you are billed until the end of the current billing cycle, and billing stops for the next billing cycle. After your storage is canceled, you might still see your storage instance in the console or the CLI for up to 72 hours.
- Immediate cancellation: Choose this option to immediately remove your storage. Neither you nor your users can use the storage anymore or recover the data.
- Anniversary date: Choose this option to cancel your storage on the next anniversary date. Your storage instances remain active until the next anniversary date and you can continue to use them until this date, such as to give your team time to make backups of your data.
-
When you dynamically provisioned the storage with a storage class that sets
reclaimPolicy: Deleteand you choose to remove the PVC, the PV and the storage instance are immediately removed. For hourly billed storage, billing stops immediately. For monthly billed storage, you are still charged for the remainder of the month. After your storage is removed and billing stops, you might still see your storage instance in the console or the CLI for up to 72 hours.
What do I need to be aware of before I delete persistent storage?
When you clean up persistent storage, you delete all the data that is stored in it. If you need a copy of the data, make a backup for file storage or block storage.
I deleted my storage instance. Why can I still see my instance?
After you remove persistent storage, it can take up to 72 hours for the removal to be fully processed and for the storage to disappear from your {{site.data.keyword.cloud_notm}} console or CLI.
{: #storage_remove}
Remove the PVC, PV, and the storage instance from your {{site.data.keyword.cloud_notm}} account to avoid further charges for your persistent storage. {: shortdesc}
Before you begin:
- Make sure that you backed up any data that you want to keep.
- Log in to your account. If applicable, target the appropriate resource group. Set the context for your cluster.
To clean up persistent data:
-
List the PVCs in your cluster and note the
NAMEof the PVC, theSTORAGECLASS, and the name of the PV that is bound to the PVC and shown asVOLUME.kubectl get pvc{: pre}
Example output:
NAME STATUS VOLUME CAPACITY ACCESSMODES STORAGECLASS AGE claim1-block-bronze Bound pvc-06886b77-102b-11e8-968a-f6612bb731fb 20Gi RWO ibmc-block-bronze 78d claim-file-bronze Bound pvc-457a2b96-fafc-11e7-8ff9-b6c8f770356c 4Gi RWX ibmc-file-bronze-retain 105d claim-file-silve Bound pvc-1efef0ba-0c48-11e8-968a-f6612bb731fb 24Gi RWX ibmc-file-silver 83d{: screen}
-
Review the
ReclaimPolicyandbillingTypefor the storage class.kubectl describe storageclass <storageclass_name>{: pre}
If the reclaim policy says
Delete, your PV and the physical storage are removed when you remove the PVC. Note that VPC Block Storage is not removed automatically, even if you used aDeletestorage class to provision the storage. If the reclaim policy saysRetain, or if you provisioned your storage without a storage class, then your PV and physical storage are not removed when you remove the PVC. You must remove the PVC, PV, and the physical storage separately.If your storage is charged monthly, you still get charged for the entire month, even if you remove the storage before the end of the billing cycle. {: important}
-
Remove any pods that mount the PVC.
-
List the pods that mount the PVC.
kubectl get pods --all-namespaces -o=jsonpath='{range .items[*]}{"\n"}{.metadata.name}{":\t"}{range .spec.volumes[*]}{.persistentVolumeClaim.claimName}{" "}{end}{end}' | grep "<pvc_name>"{: pre}
Example output:
blockdepl-12345-prz7b: claim1-block-bronze{: screen}
If no pod is returned in your CLI output, you do not have a pod that uses the PVC.
-
Remove the pod that uses the PVC. If the pod is part of a deployment, remove the deployment.
kubectl delete pod <pod_name>{: pre}
-
Verify that the pod is removed.
kubectl get pods{: pre}
-
-
Remove the PVC.
kubectl delete pvc <pvc_name>{: pre}
-
Review the status of your PV. Use the name of the PV that you retrieved earlier as
VOLUME.kubectl get pv <pv_name>{: pre}
When you remove the PVC, the PV that is bound to the PVC is released. Depending on how you provisioned your storage, your PV goes into a
Deletingstate if the PV is deleted automatically, or into aReleasedstate, if you must manually delete the PV. Note: For PVs that are automatically deleted, the status might briefly sayReleasedbefore it is deleted. Rerun the command after a few minutes to see whether the PV is removed. -
If your PV is not deleted, manually remove the PV.
kubectl delete pv <pv_name>{: pre}
-
Verify that the PV is removed.
kubectl get pv{: pre}
-
{: #sl_delete_storage}List the physical storage instance that your PV pointed to and note the
idof the physical storage instance.ibmcloud sl block volume-list --columns id --columns notes | grep <pv_name>{: pre}
Example output:
12345678 {"plugin":"ibmcloud-block-storage-plugin-689df949d6-4n9qg","region":"us-south","cluster":"aa1a11a1a11b2b2bb22b22222c3c3333","type":"Endurance","ns":"default","pvc":"mypvc","pv":"pvc-d979977d-d79d-77d9-9d7d-d7d97ddd99d7","storageclass":"ibmc-block-silver","reclaim":"Delete"}{: screen}
Understanding the Notes field information:
"plugin":"ibm-file-plugin-5b55b7b77b-55bb7": The storage plug-in that the cluster uses."region":"us-south": The region that your cluster is in."cluster":"aa1a11a1a11b2b2bb22b22222c3c3333": The cluster ID that is associated with the storage instance."type":"Endurance": The type of file or block storage, eitherEnduranceorPerformance."ns":"default": The namespace that the storage instance is deployed to."pvc":"mypvc": The name of the PVC that is associated with the storage instance."pv":"pvc-d979977d-d79d-77d9-9d7d-d7d97ddd99d7": The PV that is associated with the storage instance."storageclass":"ibmc-file-gold": The type of storage class: bronze, silver, gold, or custom.
-
Remove the physical storage instance.
ibmcloud sl block volume-cancel <classic_block_id>{: pre}
-
Verify that the physical storage instance is removed.
The deletion process might take up to 72 hours to complete. {: important}
ibmcloud sl block volume-list{: pre}