Skip to content

Latest commit

 

History

History
248 lines (208 loc) · 10.4 KB

element.rst

File metadata and controls

248 lines (208 loc) · 10.4 KB
Raw

Element (HCI/SolidFire)

To create and use an Element backend, you will need:

Preparation

The solidfire-san storage driver supports both volume modes: file and block. For the Filesystem volumeMode, Trident creates a volume and creates a filesystem. The filesystem type is specified by the StorageClass.

Driver Protocol VolumeMode Access Modes Supported File Systems Supported
solidfire-san iSCSI Block RWO,ROX,RWX No Filesystem. Raw block device
solidfire-san iSCSI Block RWO,ROX,RWX No Filesystem. Raw block device
solidfire-san iSCSI Filesystem RWO,ROX xfs, ext3, ext4
solidfire-san iSCSI Filesystem RWO,ROX xfs, ext3, ext4

All of your Kubernetes worker nodes must have the appropriate iSCSI tools installed. See the :ref:`worker configuration guide <iSCSI>` for more details.

Note

Trident will use CHAP when functioning as an enhanced CSI Provisioner.

If you're using CHAP (which is the default for CSI), no further preparation is required. It is recommended to explicitly set the UseCHAP option to use CHAP with non-CSI Trident. Otherwise, see the :ref:`access groups guide <Using access groups>` below.

Warning

Volume Access Groups are only supported by the conventional, non-CSI framework for Trident. When configured to work in CSI mode, Trident uses CHAP.

If neither AccessGroups or UseCHAP are set then one of the following rules applies: * If the default trident access group is detected then access groups are used. * If no access group is detected and Kubernetes version >= 1.7 then CHAP is used.

Backend configuration options

Parameter Description Default
version Always 1  
storageDriverName Always "solidfire-san"  
backendName Custom name for the storage backend "solidfire_" + storage (iSCSI) IP address
Endpoint MVIP for the SolidFire cluster with tenant credentials  
SVIP Storage (iSCSI) IP address and port  
labels Set of arbitrary JSON-formatted labels to apply on volumes. ""
TenantName Tenant name to use (created if not found)  
InitiatorIFace Restrict iSCSI traffic to a specific host interface "default"
UseCHAP Use CHAP to authenticate iSCSI  
AccessGroups List of Access Group IDs to use Finds the ID of an access group named "trident"
Types QoS specifications (see below)  
limitVolumeSize Fail provisioning if requested volume size is above this value "" (not enforced by default)
debugTraceFlags Debug flags to use when troubleshooting. E.g.: {"api":false, "method":true} null

Warning

Do not use debugTraceFlags unless you are troubleshooting and require a detailed log dump.

Note

For all volumes created, Trident will copy all labels present on a storage pool to the backing storage LUN at the time it is provisioned. Storage admins can define labels per storage pool and group all volumes created in a storage pool. This provides a convenient way of differentiating volumes based on a set of customizable labels that are provided in the backend configuration.

Example configuration

Example 1 - Backend configuration for solidfire-san driver with three volume types

This example shows a backend file using CHAP authentication and modeling three volume types with specific QoS guarantees. Most likely you would then define storage classes to consume each of these using the IOPS storage class parameter.

{
    "version": 1,
    "storageDriverName": "solidfire-san",
    "Endpoint": "https://<user>:<password>@<mvip>/json-rpc/8.0",
    "SVIP": "<svip>:3260",
    "TenantName": "<tenant>",
    "labels": {"k8scluster": "dev1", "backend": "dev1-element-cluster"},
    "UseCHAP": true,
    "Types": [{"Type": "Bronze", "Qos": {"minIOPS": 1000, "maxIOPS": 2000, "burstIOPS": 4000}},
              {"Type": "Silver", "Qos": {"minIOPS": 4000, "maxIOPS": 6000, "burstIOPS": 8000}},
              {"Type": "Gold", "Qos": {"minIOPS": 6000, "maxIOPS": 8000, "burstIOPS": 10000}}]
}

Example 2 - Backend and Storage Class configuration for solidfire-san driver with Virtual Storage Pools

This example shows the backend definition file configured with :ref:`Virtual Storage Pools <Virtual Storage Pools>` along with StorageClasses that refer back to them.

In the sample backend definition file shown below, specific defaults are set for all storage pools, which set the type at Silver. The Virtual Storage Pools are defined in the storage section. In this example, some of the storage pool sets their own type, and some pools overwrite the default values set above.

{
    "version": 1,
    "storageDriverName": "solidfire-san",
    "Endpoint": "https://<user>:<password>@<mvip>/json-rpc/8.0",
    "SVIP": "<svip>:3260",
    "TenantName": "<tenant>",
    "UseCHAP": true,
    "Types": [{"Type": "Bronze", "Qos": {"minIOPS": 1000, "maxIOPS": 2000, "burstIOPS": 4000}},
              {"Type": "Silver", "Qos": {"minIOPS": 4000, "maxIOPS": 6000, "burstIOPS": 8000}},
              {"Type": "Gold", "Qos": {"minIOPS": 6000, "maxIOPS": 8000, "burstIOPS": 10000}}],

    "type": "Silver",
    "labels":{"store":"solidfire", "k8scluster": "dev-1-cluster"},
    "region": "us-east-1",

    "storage": [
        {
            "labels":{"performance":"gold", "cost":"4"},
            "zone":"us-east-1a",
            "type":"Gold"
        },
        {
            "labels":{"performance":"silver", "cost":"3"},
            "zone":"us-east-1b",
            "type":"Silver"
        },
        {
            "labels":{"performance":"bronze", "cost":"2"},
            "zone":"us-east-1c",
            "type":"Bronze"
        },
        {
            "labels":{"performance":"silver", "cost":"1"},
            "zone":"us-east-1d"
        }
    ]
}

The following StorageClass definitions refer to the above Virtual Storage Pools. Using the parameters.selector field, each StorageClass calls out which virtual pool(s) may be used to host a volume. The volume will have the aspects defined in the chosen virtual pool.

The first StorageClass (solidfire-gold-four) will map to the first Virtual Storage Pool. This is the only pool offering gold performance with a Volume Type QoS of Gold. The last StorageClass (solidfire-silver) calls out any storage pool which offers a silver performance. Trident will decide which Virtual Storage Pool is selected and will ensure the storage requirement is met.

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: solidfire-gold-four
provisioner: csi.trident.netapp.io
parameters:
  selector: "performance=gold; cost=4"
  fsType: "ext4"
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: solidfire-silver-three
provisioner: csi.trident.netapp.io
parameters:
  selector: "performance=silver; cost=3"
  fsType: "ext4"
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: solidfire-bronze-two
provisioner: csi.trident.netapp.io
parameters:
  selector: "performance=bronze; cost=2"
  fsType: "ext4"
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: solidfire-silver-one
provisioner: csi.trident.netapp.io
parameters:
  selector: "performance=silver; cost=1"
  fsType: "ext4"
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: solidfire-silver
provisioner: csi.trident.netapp.io
parameters:
  selector: "performance=silver"
  fsType: "ext4"

Using access groups

Note

Ignore this section if you are using CHAP, which we recommend to simplify management and avoid the scaling limit described below. In addition, if using Trident in CSI mode, you can safely ignore this section. Trident uses CHAP when installed as an enhanced CSI provisioner.

Trident can use volume access groups to control access to the volumes that it provisions. If CHAP is disabled it expects to find an access group called trident unless one or more access group IDs are specified in the configuration.

While Trident associates new volumes with the configured access group(s), it does not create or otherwise manage access groups themselves. The access group(s) must exist before the storage backend is added to Trident, and they need to contain the iSCSI IQNs from every node in the Kubernetes cluster that could potentially mount the volumes provisioned by that backend. In most installations that's every worker node in the cluster.

For Kubernetes clusters with more than 64 nodes, you will need to use multiple access groups. Each access group may contain up to 64 IQNs, and each volume can belong to 4 access groups. With the maximum 4 access groups configured, any node in a cluster up to 256 nodes in size will be able to access any volume.

If you're modifying the configuration from one that is using the default trident access group to one that uses others as well, include the ID for the trident access group in the list.