Skip to main content

Add VM Disks

Purpose: For operators, shows how to attach additional storage volumes with provider-specific procedures.

When to Add Disks​

Common reasons to attach additional disks to cluster nodes:

  • Longhorn storage pool needs more capacity
  • etcd data directory is running low on space (control plane nodes)
  • A workload requires dedicated local storage (e.g., database nodes)
  • Container runtime storage (/var/lib/containerd) is filling up

OpenStack: Attach a Cinder Volume​

Step 1: Create and Attach the Volume via Terraform​

# infrastructure/clusters/<cluster>/main.tf
resource "openstack_blockstorage_volume_v3" "data_disk" {
  name        = "worker-01-data"
  size        = 500  # GB
  volume_type = "ssd"
}

resource "openstack_compute_volume_attach_v2" "worker_01_data" {
  instance_id = openstack_compute_instance_v2.workers[0].id
  volume_id   = openstack_blockstorage_volume_v3.data_disk.id
  device       = "/dev/vdb"
}
cd infrastructure/clusters/<cluster>/
terraform plan
terraform apply

Step 2: Format and Mount on the Node​

SSH to the node and prepare the disk:

# Verify the disk is visible
lsblk

# Create a filesystem
sudo mkfs.ext4 /dev/vdb

# Create mount point and mount
sudo mkdir -p /mnt/data
sudo mount /dev/vdb /mnt/data

# Add to fstab for persistence across reboots
echo '/dev/vdb /mnt/data ext4 defaults 0 2' | sudo tee -a /etc/fstab

Step 3: Configure Longhorn (if used for storage pool)​

If the disk is for Longhorn, add it as a storage path on the node:

# Create the Longhorn data directory on the new disk
sudo mkdir -p /mnt/data/longhorn

Then update the Longhorn node configuration:

apiVersion: longhorn.io/v1beta2
kind: Node
metadata:
  name: worker-01
  namespace: longhorn-system
spec:
  disks:
    default-disk:
      path: /var/lib/longhorn
      allowScheduling: true
    data-disk:
      path: /mnt/data/longhorn
      allowScheduling: true
      storageReserved: 10737418240  # 10 GB reserved

VMware: Add a Virtual Disk​

Step 1: Add the Disk via Terraform​

# infrastructure/clusters/<cluster>/main.tf
resource "vsphere_virtual_machine" "worker_01" {
  # ... existing configuration ...

  disk {
    label            = "disk0"
    size             = 100
    unit_number      = 0
  }

  # Additional data disk
  disk {
    label            = "disk1"
    size             = 500
    unit_number      = 1
    thin_provisioned = true
  }
}
terraform plan
terraform apply

VMware hot-adds the disk without requiring a VM restart (if the VM hardware version supports it).

Step 2: Format and Mount​

Follow the same format and mount steps as OpenStack (the disk appears as /dev/sdb on VMware).

Verification​

# Confirm the disk is mounted
df -h | grep /mnt/data

# Check Longhorn disk status (if applicable)
kubectl get nodes.longhorn.io -n longhorn-system -o yaml | grep -A10 disks

# Verify no I/O errors
dmesg | grep -i error | grep -i disk

Troubleshooting​

  • Disk not visible after attach — Rescan the SCSI bus: echo "- - -" | sudo tee /sys/class/scsi_host/host*/scan. On VMware, the guest OS may need a reboot if hot-add is not supported.
  • Longhorn not using new disk — Verify the Node CR has the new disk path and allowScheduling: true. Check Longhorn manager logs.
  • fstab entry causes boot failure — Use nofail mount option: /dev/vdb /mnt/data ext4 defaults,nofail 0 2. This prevents boot hangs if the disk is detached.