Isilon OneFS: Virtual Cluster Deployment - Part 1: Installation & Initial Configuration

Welcome to the first installment of the Isilon / PowerScale series. In this documentation, we will explore the depths of Dell EMC’s scale-out NAS platform, starting from the ground up. Whether you are building a lab for testing or preparing for an enterprise deployment, this guide will provide the technical foundations needed to stand up a virtual Isilon cluster on Proxmox.

Phase 1: MAC, VLAN, and VM Configuration in Proxmox

Create a VM without disks, with 3 Network Interface Cards (NICs), a minimum of 6GB of RAM, and 4 vCPUs. Edit the hardware for each cloned node to match your network design exactly.

Your VM should look like this:

Assigning static MAC addresses and separating traffic using VLAN tags directly on the bridge (vmbr0) is the best way to maintain order and avoid identification headaches within OneFS.

Phase 2: OVA and Disk Import

Since you already have the hardware structure ready in VM 201, the correct flow is to import the OVA and then ensure the data disks are attached.

1. Import the OVA

If you haven’t already, extract the OVA and import the system disk (.ovf/.vmdk). This will create the node’s main (OS) disk.

Navigate to the path where the OVA is located:

# Extract OVA contents
tar -xvf Isilon_9.7.0.0.ova

Create a file named import_isilon.sh, give it execution permissions (chmod +x import_isilon.sh), and run it in the same folder where you extracted the OVA. Adapt the VMID and Storage depending on where you want to store the disks.

#!/bin/bash

# Configuration variables
VMID=201
STORAGE="Black_Box_SSD_02"

# Loop through the 22 disks from the Isilon OVA
for i in {1..22}; do
    FILE="9.7.0.0-disk${i}.vmdk"

    # Check if the VMDK file exists
    if [ ! -f "$FILE" ]; then
        echo "Warning: $FILE not found, skipping..."
        continue
    fi

    echo "Importing $FILE to $STORAGE for VM $VMID..."
    
    # Convert and import the disk to the specified storage
    qm importdisk $VMID "$FILE" $STORAGE --format qcow2

    # Find the newly created unused disk name
    UNUSED_DISK=$(qm config $VMID | grep "^unused" | head -n 1 | awk '{print $2}')

    if [ -z "$UNUSED_DISK" ]; then
        echo "Error: Could not find unused disk for $FILE"
        exit 1
    fi

    # SCSI index (Disk 1 -> scsi0, Disk 2 -> scsi1, etc.)
    SCSI_IDX=$((i - 1))
    
    echo "Attaching $UNUSED_DISK as scsi${SCSI_IDX}..."
    
    # Attach disk to SCSI bus
    qm set $VMID --scsi${SCSI_IDX} "$UNUSED_DISK"
done

echo "Process complete! All 22 disks successfully attached as SCSI to VM $VMID."

2. Verify Disks in Proxmox

Once the script finishes importing, you can check the status of the disks in your Proxmox.

3. Clone for Multi-Node Cluster

Before powering on VM 201, perform a Full Clone twice to create Node 2 and Node 3.

Once you have the 3 VMs, continue with the first VM.

Phase 3: Node 1 Initialization (Cluster Creation)

Power on Node 1, open the VNC console, and wait for the configuration wizard to load. At startup, it will ask to format all disks (ifs partition); press yes.

1. Cluster Selection

Select Create a new cluster (option 1).

2. EULA and Credentials

Accept the EULA and define passwords for root and admin.

3. Cluster Naming

Name your cluster (e.g., Isilon-Lab).

4. Encoding

Select the cluster encoding; use the default.

5. Int-a Configuration (Backend)

• Interface: Look for the one that matches MAC 02:00:00:AA:CC:01 (likely listed as vtnet2).
• Netmask: 255.255.255.0
• IP Range: Use a fictional range, e.g., 10.10.10.1 to 10.10.10.10. This traffic will live only within the isolated VLAN 1000.

6. Int-b / Failover Configuration

Skip it (leave blank/Enter).

7. Ext-1 Configuration (MGMT)

• Interface: Matching MAC 02:00:00:AA:AA:01 (likely vtnet0).
• Subnet: 255.255.255.0
• IP Range: Assign a range from your 10.0.6.100/24 network (e.g., 10.0.6.21 to 10.0.6.125).

• Gateway: The IP of your default gateway for the 10.0.6.0 network.

Phase 4: DNS and Timezone Configuration

Upon pressing Enter, the wizard will ask for final infrastructure parameters:

• DNS Servers: Enter the IP of your local DNS server (if you have Active Directory/Pi-hole in your lab) or a public one like 1.1.1.1.
• Search Domains: You can leave this blank or use your lab domain (e.g., lab.local).
• Timezone: Select your region (e.g., America/Tijuana) so cluster logs match your local time.

Join Mode Selection

Define the security behavior required when Node 2 and 3 join:

• Manual (Recommended): Any new node that starts and detects this cluster via the backend network can join if you select the option in its wizard. Fixed for lab environments.
• Secure: Requires an additional layer of authentication.

Press Enter to keep the default option (Manual). Type yes and press Enter to commit the changes. The node will begin formatting the 21 SCSI disks and building the /ifs file system.

To log in, go to: https://10.0.6.100:8080/OneFS/Login

You should be able to see the web interface now.

Conclusion

With this, the first node of our virtual Isilon cluster is live and accessible via the web interface. In the next installment of this series, we will configure the joining of the remaining nodes and prepare the data network for SMB services.

End of transmission.