Skip to main content

Beta Release Version v2.2.1

· 6 min read

We're excited to announce Version v2.2.1 of the Hypervisor Control Panel! This release completely rebuilds the instance backup experience with support for remote backup destinations and adds GPU passthrough for compute-heavy workloads.

You can now back up your instances directly to Amazon S3, MinIO, Wasabi, Backblaze B2, DigitalOcean Spaces, SFTP servers, FTP servers, or any rclone-supported backend -- no need to mount backup storage locally on each hypervisor.

  • [Feature] Remote Backup Destinations - Configure S3, SFTP, FTP, or rclone-based backup targets from the admin panel. Credentials are stored encrypted and pushed to hypervisors only when needed.
  • [Feature] Streaming Backups - Full and incremental backups stream directly to the remote destination. Restores stream back the same way. No temporary disk space required.
  • [Feature] Running-VM Backups Without Downtime - Backups of running VMs use external snapshots so the VM continues running normally while the backup completes. Changes made during the backup window are merged back seamlessly.
  • [Feature] Efficient Incremental Backups - Incremental backups capture only the blocks that changed since the last backup, dramatically reducing backup size and duration.
  • [Feature] Parallel Backups - Multiple backups can now run concurrently on the same hypervisor without conflict.
  • [Feature] GPU Passthrough - PCIe GPU passthrough for compute-heavy workloads like ML training, rendering, and video encoding. Hypervisors automatically discover installed GPUs and show availability in the admin panel.
  • [Feature] GPU Instance Plans - Instance plans can now specify a GPU count, enabling GPU-backed instance deployment through the standard creation flow.
  • [Improvement] Redesigned AI Assistant - Cleaner sidebar chat panel with smooth transitions, a welcome screen with suggested prompts, and proper mobile sizing.
  • [Improvement] Multi-Distro Docker Support - The Docker installation option during instance deploy now works on Ubuntu, Debian, AlmaLinux, Rocky Linux, and CentOS. RHEL-based distros automatically get SELinux disabled to prevent container permission issues.
  • [Improvement] Backup Scheduling via Policies - All automated backup scheduling now happens through backup policies. The legacy per-instance backup settings have been retired in favor of the more flexible policy-based approach.
  • [Improvement] Pending Invitation Login Protection - Users with unaccepted team invitations can no longer log in or reset their password until they accept the invitation, eliminating confusion around unaccepted accounts.
  • [Security] Password Reset Hardening - The forgot-password endpoint now returns a uniform response for all cases, preventing attackers from discovering whether an email address has an account.
  • [Security] Encrypted Backup Credentials - S3 keys, SFTP passwords, and other sensitive backup configuration are stored encrypted.
  • [Fix] Scaling Group Modals - Fixed dropdowns in the Change Load Balancer and related modals not showing the selected value correctly.
  • [Fix] Instance Network Tab - Adding, removing, or setting an IP as primary now immediately updates the IP list without requiring a page refresh.
  • [Fix] Accept Invitation Form - Confirm password field moved above the timezone selector for natural field ordering.
  • [Fix] GPU Devices Card Layout - Fixed spacing issue where the GPU Devices section stuck to the next card on the hypervisor detail page.
  • [Fix] Instance Plan Badges - Datastore type (Ceph, etc.) now renders as a proper badge alongside the Datastore label.

Streaming Backups​

The backup system has been rebuilt to give you flexibility in where your backups are stored. Previously, each hypervisor needed a local mount point for backups -- whether NFS, SMB, or local disk. Now, backups can stream directly to any of the following destinations:

  • Amazon S3 and any S3-compatible service (MinIO, Wasabi, Backblaze B2, DigitalOcean Spaces, etc.)
  • SFTP servers
  • FTP servers
  • Any rclone remote (40+ backends supported, including Google Cloud Storage, Azure Blob, and more)
  • Local mount points (backward compatible with existing setups)

What this means for you​

  • No more local disk planning for backups -- back up straight to your existing cloud storage
  • Offsite backups by default -- your backup data lives in a different location from your VMs
  • Better cost control -- use cheap object storage tiers like Backblaze B2 or Wasabi for long-term backup retention
  • Simpler hypervisor setup -- no NFS exports or SMB mounts to manage

How it works​

Backups of running VMs complete without any downtime. Your VM continues running normally throughout the backup window. Behind the scenes, we take an instant snapshot of the disk so the backup reads a consistent point-in-time view while your VM keeps writing to a temporary overlay. When the backup finishes, the overlay is merged back seamlessly.

Incremental backups track changed blocks since the last backup, so subsequent backups are much smaller and faster than the initial full backup. Restores automatically walk the chain from the full backup through all incrementals to rebuild your data exactly as it was.

Configuring backup storage​

Admins can create new backup storage targets from the Backup Storage page in the admin panel. Simply select the storage type (S3, SFTP, FTP, rclone, or Local), fill in the connection details, and enable it. All credentials are stored encrypted.

See the full Instance Backups feature guide for setup walkthroughs for each storage type.

GPU Passthrough​

PCIe GPU passthrough is now supported for workloads that need direct GPU access -- ML training, AI inference, rendering, video encoding, and more.

When you provision a hypervisor with GPUs installed, they are automatically discovered and cataloged. The admin hypervisor detail page shows:

  • All installed GPUs with model and VRAM
  • Current allocation status (available, allocated, in-maintenance)
  • Which instance each allocated GPU is assigned to
  • NUMA affinity information

Instance plans can specify a GPU count, so you can offer GPU-backed plans to your customers. When a customer deploys an instance with such a plan, the platform automatically allocates an available GPU from the target hypervisor and attaches it to the VM.

NVIDIA Multi-Instance GPU (MIG) partitions on A100/H100 GPUs and SR-IOV virtual functions are detected automatically.

Backup Policy Migration​

With backup policies handling all scheduled backups, we've retired the older per-instance backup scheduling UI. The "Update Backup Settings" option on the admin instance page has been removed.

All automated backups now run through Backup Policies, which give you:

  • Centralized scheduling across many instances with one policy
  • Configurable full and incremental frequencies
  • Retention control
  • Failure notification and auto-pause on repeated failures

Manual "Create Backup" buttons on the instance Backups tab remain available for one-off backups.

Migration Guide​

  1. Update the control panel and re-provision your hypervisors so they pick up the new backup tooling.
  2. Existing backup storage entries continue working as "Local" type without any changes.
  3. Create new backup storage entries of type S3/SFTP/FTP/rclone from the Backup Storage admin page.
  4. If you were relying on the legacy per-instance backup settings, create a backup policy and attach your instances to it.

Full feature guide: Instance Backups