1.5.0

We are excited to announce our release for aleph-vm in version 1.5.0!

This new release of aleph-vm brings significant improvements to the Aleph VM, including the new GPU reservation system and in-VM reboot functionality. It also addresses several important bug fixes related to CI stability, GPU detection, network configuration, VM lifecycle, and volume downloads. The upgrade of Firecracker and aleph-message, along with the Pydantic migration, further enhances the platform.

New Features

• GPU Reservation System: Implemented a system for reserving GPUs. (PR #776)
• VM Reboot from Inside: Allowed users to reboot their VMs from within the VM itself. (PR #790)
• Operator Authentication Improvements: Enhanced operator authentication with better HTTP error reporting. (PR #788)

Bug Fixes

• Random CI Failures: Fixed random failures in the Continuous Integration (CI) pipeline and merged workflows. (PR #787)
• GPU x-vga Detection: Corrected the detection of x-vga for GPUs. (PR #786)
• ndppd Proxy Interface Loading: Fixed an issue where the ndppd proxy was not loading existing interfaces on supervisor restart. (PR #783)
• VM Lifecycle and Blocking Resources: Addressed issues related to the VM lifecycle and blocking resources. (PR #789)
• Volume Download from Connector: Fixed a problem with downloading volumes from the connector. (PR #796)
• Pydantic TypeAdapter Issue: Solved an issue related to Pydantic's TypeAdapter. (PR #794)
• Confidential VM Cleanup (ALEPH-517): Fixed an issue related to the cleanup of confidential VMs. (PR #797)
• Confidential and GPU Instance Stopping (ALEPH-518): Resolved a bug where confidential and GPU instances were stopping immediately after starting. (PR #798)
• Message Requirements Casting: Solved a casting issue with message requirements. (PR #800)
• Internet Check Failure: Fixed an issue causing the internet check to fail. (PR #799)

Improvements

• Hatch Formatting: Performed some cleanup using hatch fmt. (PR #782)
• Usage System: Removed reserved usage tracking for volumes in the usage system. (PR #784)
• Misc Stability Improvements: Implemented various stability improvements (OL ALEPH-166). (PR #719)
• Firecracker Upgrade: Upgraded Firecracker from version 1.5.0 to 1.7.0. (PR #771)
• Pydantic Migration: Migrated the project from Pydantic V1 to V2. (PR #791)
• Confidential VM Locale: Updated locale settings to en_US UTF-8 during the setup of confidential VMs. (PR #792)
• aleph-message Upgrade: Upgraded the aleph-message dependency to version 1.0.0. (PR #795)

What's Changed

• CI: Fix random failures. Merge workflows by @olethanh in #787
• Bit of clean up with hatch fmt by @olethanh in #782
• Gpu Reservation system by @olethanh in #776
• Fix x-vga detection on GPUs by @nesitor in #786
• Fix: ndppd proxy not loading existing interfaces on supervisor restart by @nesitor in #783
• Usage system: remove reserved usage for volumes by @olethanh in #784
• Ol aleph 166 misc stability by @olethanh in #719
• Fix: Upgrade Firecracker 1.5.0 -> 1.7.0 by @hoh in #771
• Fix VM lifecycle and blocking resource. by @olethanh in #789
• enh: allow user to reboot from inside the VM by @olethanh in #790
• enh: Operator auth. Better http error by @olethanh in #788
• Migrate Pydantic V1 to V2 by @nesitor in #791
• Solve Pydantic TypeAdapter issue by @nesitor in #794
• Confidentials: Update locale settings to en_US UTF-8 during setup by @aliel in #792
• FIX downloading volume from connector by @olethanh in #796
• Upgrade aleph-message==1.0.0 version dependency by @nesitor in #795
• Fix ALPEH-517 confidential clean up by @olethanh in #797
• ALEPH-518 Fix Confidential and GPU instance stopping after start by @olethanh in #798
• Solve message requirements casting issue by @nesitor in #800
• Fix internet check failure by @olethanh in #799

Full Changelog: 1.4.2...1.5.0

How to upgrade

1. Upgrade the packages

This part did not change, download and install the new package as usual.

On Debian 12 (Bookworm):

rm -f /opt/aleph-vm.debian-12.deb
wget -P /opt https://github.com/aleph-im/aleph-vm/releases/download/1.5.0/aleph-vm.debian-12.deb
apt install /opt/aleph-vm.debian-12.deb

On Ubuntu 22.04 (Jammy Jellyfish):

sudo rm -f /opt/aleph-vm.ubuntu-22.04.deb
sudo wget -P /opt https://github.com/aleph-im/aleph-vm/releases/download/1.5.0/aleph-vm.ubuntu-22.04.deb
sudo apt install /opt/aleph-vm.ubuntu-22.04.deb

On Ubuntu 24.04 (Noble Numbat):

sudo rm -f /opt/aleph-vm.ubuntu-24.04.deb
sudo wget -P /opt https://github.com/aleph-im/aleph-vm/releases/download/1.5.0/aleph-vm.ubuntu-24.04.deb
sudo apt install /opt/aleph-vm.ubuntu-24.04.deb

2. Enable GPU support (optional)

In order to enable GPU support on your Compute Resource Node, you must:

1. Ensure that your system have a compatible GPU card.
2. Detach GPU cards from the kernel module drivers and attach it to QEMU vfio drivers.
3. Enable GPU support in the aleph-vm configuration.

Please follow these instructions

Enable GPU in the configuration of aleph-vm, by default in /etc/aleph-vm/supervisor.env. This is not the default yet.

ALEPH_VM_ENABLE_GPU_SUPPORT=True

After launching the server you can check the endpoint
http://localhost:4020/status/config or https://<your-node-domain>/status/config and verify that ENABLE_GPU_SUPPORT has the value true.