Skip to main content
Crusoe Support Help Center home page
Crusoe

How-To Safely Upgrade the VAST NFS Driver on a Crusoe VM

Sagar Lulla
Sagar Lulla
Updated

Introduction

This guide shows you how to check the version of the VAST NFS client driver (vastnfs) on a Crusoe VM and upgrade it safely without corrupting active mounts.

Crusoe shared disks are served over NFSv3 from a VAST storage cluster. The high-throughput mount options used with shared disks — nconnect, remoteports, spread_reads, spread_writes — are provided by the vastnfs kernel driver rather than the stock in-tree NFS client.

The driver ships preinstalled in Crusoe curated images and replaces the standard NFS kernel modules, so lsmod output looks identical to a vanilla client. The reliable way to identify the installed version is vastnfs-ctl status.

Because the driver consists of kernel modules, it cannot be reloaded while any NFS filesystem is mounted. Upgrading therefore requires briefly unmounting all NFS volumes, or rebooting — which is why this guide is written around a maintenance window rather than a live workload.

Prerequisites

  • Sudo Access on the VM
  • Maintenance Window Allowing All NFS Mounts to Be Unmounted or a VM Reboot
  • Outbound Internet Access From the VM (Direct Download Only)
  • No Active Workloads Writing to NFS-Backed Paths During the Upgrade

Instructions

Step 1: Check the Current Driver Version

vastnfs-ctl status

Example output:

version: 4.5.1-vastdata
kernel modules: sunrpc rpcrdma compat_nfs_ssc lockd nfs_acl nfs nfsv3
services: rpcbind.socket rpcbind
rpc_pipefs: /run/rpc_pipefs

The version line is the installed driver version. The current package published for Crusoe VMs is 4.5.8; if vastnfs-ctl status reports anything lower, an upgrade is available.

Step 2: Download the Updated Driver Package

Download the vastnfs-dkms package on the VM:

wget https://github.com/crusoecloud/crusoe-nfs-support/raw/refs/heads/main/debs/vastnfs-dkms_4.5.8-vastdata_all.deb

Confirm the download is an actual package rather than an error page saved under the same filename:

file vastnfs-dkms_4.5.8-vastdata_all.deb

It should report Debian binary package. If it reports HTML or text, the download failed and the file is an error page — apt install would fail later with a confusing message.

ℹ️ Note: If your VM has no outbound internet access, download the package on a machine that does and copy it to the VM with scp.

Step 3: Unmount All NFS Volumes

The kernel modules cannot be reloaded while any NFS mount is active. List active NFS mounts and unmount each:

findmnt -t nfs
sudo umount <each-nfs-mountpoint>
findmnt -t nfs

The second findmnt -t nfs should return nothing.

⚠️ Warning: If your home directory or other login-critical paths are NFS-mounted, unmounting them will disrupt active sessions. In that case, skip this step, install the package in Step 4, and reboot instead of reloading live — the result is the same.

Step 4: Install the Package and Reload the Driver

sudo apt install ./vastnfs-dkms_4.5.8-vastdata_all.deb
sudo vastnfs-ctl reload

DKMS builds the modules against your running kernel during installation. If vastnfs-ctl reload reports modules still in use, reboot the VM instead:

sudo reboot

Step 5: Verify the New Version and Remount

vastnfs-ctl status
sudo mount -a
findmnt -t nfs

vastnfs-ctl status should now report the upgraded version — for example version: 4.5.8-vastdata — and your fstab-defined NFS mounts should be active again with the expected options.

Confirm a write on a read-write volume:

touch <your-rw-mountpoint>/writetest && rm <your-rw-mountpoint>/writetest

Example

An engineer maintains a fleet of VMs mounting shared datasets from VAST. After a client-side driver fix is released, they schedule a 15-minute window per VM.

They check vastnfs-ctl status (4.5.1), install the 4.5.8 package, and reboot rather than reloading live — their /home is NFS-mounted, so unmounting would drop their own session. After the reboot, vastnfs-ctl status reports 4.5.8 and mount -a has restored every fstab mount. Total downtime per VM is one reboot cycle.

Related Articles

Additional Resources

Related to

Was this article helpful?

0 out of 0 found this helpful

Still need help?

Our support team is ready to assist you with any questions.

Have more questions? Submit a request

Related Articles

Recently Viewed

Comments

0 comments

Article is closed for comments.