Configuring hosts with NVMe-oF

Prerequisites

To configure NVMe-oF with Active Cluster within your environment please ensure following targets are met:

Purity//FA: version 6.10.5 or higher
VCF / vSphere: VCF 9.1, ESXi 9.1 hosts
Protocol: NVMe‑TCP or NVMe-FC
Datastore type: VMFS (stretched, ActiveCluster)
Usage in VCF: External/supplemental storage for workload domains

Configure NVMe‑oF AC host objects

A FlashArray host object is a collection of a host's initiators that can be "connected" to a volume. This allows those specified initiators (and therefore that host) to access that volume or volumes.

Create a host object on the FlashArray by going to the Storage section and then the Hosts tab.

Hosts

Click on the plus sign in the Hosts panel to create a new host. Assign the host a name that makes sense and click Create.
Obtain a unique NQN identifier from your ESXi host using ESXi CLI command
```
esxcli nvme info get
```
Or you can copy NQN using vSphere Client, go to your ESXi host → Configure → Storage Adapters -> for NVMe-TCP select VMware NVMe over TCP Storage Adapter or for NVMe-FC select your NVMe capable Fibre Channel HBA adapter → click Add Controller under Controllers tab (do not new add controller, only copy Host NQN from here).
Click on the newly created host and then in the Host Ports panel, click the vertical ellipsis and choose Configure NQNs.
Paste ESXi host NQN and click Add.
Repeat these steps for all ESXi hosts in your Active Cluster; all hosts in the ESXi cluster should be configured on both FlashArrays and added to their host group.

Set Host Personality and nvme-stretch mode

It’s required to set correct host personality and configure nvme-stretch mode for each host which will participate in NVMe-oF Active Cluster.

Important: In Purity/FA version 6.10.x, it's not possible to set host personality to ESXi 9.1 via GUI, please use CLI for setting host personality.

nvme-stretch modes

Mode	Default	Can connect to stretched pods	Subsystem type	Typical use case
disallow	yes	No, connections to stretched pods are blocked	Fleet-independent NVMe subsystem	Host is disabled for NVMe-oF Active Cluster
allow	no	Yes	Fleet-independent NVMe subsystem	NVMe-oF AC hosts which are not part of a fleet or don't need Direct Access*
allow-only-with-fleet	no	Yes, within a correctly configured fleet only	Fleet-aware subsystem	Hosts using Direct Access* across multiple arrays in a fleet

* Direct Access - the fleet-aware subsystem NQN presented by the arrays is derived from the host's own NQN, so the host sees a single, consistent NVMe subsystem across all participating arrays.

SSH to FlashArray CLI with a user who has array_admin permissions.
For each host which will participate in NVMe-oF Active Cluster set personality to ESXi9.1 using following command:
```
purehost setattr --personality esxi9.1 <HOST_NAME>
```
For each host which will participate in NVMe-oF Active Cluster configure nvme-stretched mode using following command:
```
purehost setattr --nvme-stretch allow <HOST_NAME>
```

Volumes

The first step is to create a volume on FlashArray in site A, add it to previously created stretched pod and connect the volume to your hosts or host group (if configured).

Important: Current version of Everpure Plugin for the vSphere Client does not support creating Active Cluster NVMe-oF volumes. Please don’t use the plugin for creating volumes and configure volumes manually via Purity//FA GUI or CLI.

Click on the plus sign in the Volumes panel to create a new volume.

After creating a volume and attaching it to a stretched pod, the volume should be visible within Pods under FlashArray in site B. Next step is to connect this volume to your hosts or host group (if configured) on the FlashArray in site B.

Now you can verify that the previously created volume is visible to ESXi hosts.

Open your vSphere Client, go to your ESXi host → Configure → Storage Adapters → for NVMe-TCP select VMware NVMe over TCP Storage Adapter or for NVMe-FC select your NVMe capable Fibre Channel HBA adapter → click Devices tab and confirm the volume is listed here.

Create a VMFS Datastore

Important: Current version of Everpure Plugin for the vSphere Client does not support creating Active Cluster NVMe-oF datastores. Please don't use the plugin for creating a datastore and create a datastore manually via vSphere Client.

To create a datastore manually, right click on your cluster, select Storage and click on the New Datastore.

The New Datastore wizard window will open, select VMFS as a type of the datastore and click Next button.

On the next window specify the name of the datastore, choose one of your ESXi hosts listed here and select the volume we created in the previous step.

Review the disk layout and confirm or adjust the size of the datastore.

After the datastore is created it should be automatically mounted to all ESXi hosts in the cluster. You can verify it in the vSphere Client by going to Storage → select Datastore → click on Hosts tab. Confirm that all hosts in the cluster are successfully connected to this datastore.

Important: If your environment is configured as non-uniform Active Cluster with two vCenter servers (one per site), do Rescan Storage on your ESXi hosts on secondary site to see newly created datastore.