The NVIDIA DGX Station handles upgrades to its major releases through a structured process that involves updating package sources and installing new software versions. Here's a detailed overview of how these upgrades are managed:
Prerequisites for Major Release Upgrades
Before upgrading to a new major release, ensure that your current DGX OS Desktop release is on a supported update path for the target release. You must also be logged in as an administrator user on the DGX Station[1].
Steps for Upgrading to a New Major Release
1. Update Package Information: Start by downloading information from all configured sources about the latest versions of packages using the command:
bash
sudo apt update
2. Install Available Updates: Install all available updates for your current DGX OS Desktop release to ensure your system is up-to-date:
bash
sudo apt -y full-upgrade
3. Install the Upgrade Package: Install the `dgx-release-upgrade` package, which is necessary for upgrading to a new major release:
bash
sudo apt install -y dgx-release-upgrade
4. Initiate the Upgrade Process: Start the DGX OS Desktop release upgrade process using the following command:
bash
sudo dgx-release-upgrade
5. Handle Remote SSH Sessions: If you are upgrading remotely via SSH, you will be prompted to continue running under SSH. It is recommended to avoid upgrading over SSH due to potential recovery issues in case of failure. However, if you choose to proceed, an additional SSH daemon will be started on port 1022 to facilitate recovery[1].
6. Review and Remove Obsolete Packages: During the upgrade, you may be prompted to review and remove obsolete packages. Check the lists of obsolete packages in the DGX OS Desktop Release Notes to ensure you are removing only unwanted packages[1].
7. Complete the Upgrade and Restart: Once the upgrade is complete, you will be prompted to restart the system. It is crucial to restart the DGX Station to finalize the updates, especially if NVIDIA Graphics Drivers were updated[1].
Post-Upgrade Verification
After the system restarts, verify that the upgrade was successful by checking the Linux kernel version and NVIDIA Graphics Drivers version using commands like `uname -a` and `nvidia-smi`, respectively[1].
Additional Considerations
- Security Updates: Keep your system updated with the latest security patches to protect against vulnerabilities. Check Ubuntu Security Notices for relevant updates[1].
- Patch Updates: To receive patch updates, you must opt-in by installing the appropriate package (e.g., `dgxstation-release-updates-repo`) and updating your package sources[1].
Overall, upgrading to a new major release on the DGX Station involves updating package sources, installing necessary packages, and ensuring compatibility with the new release. It's essential to follow the recommended steps carefully to avoid potential issues during the upgrade process.
Citations:
[1] http://cdn.cnetcontent.com/2f/ed/2fed407b-03de-4bda-86b4-64cb9ec20a50.pdf
[2] https://kb.brightcomputing.com/knowledge-base/installing-nvidia-dgx-software-stack-in-bright-rhel8-software-images/
[3] https://docs.nvidia.com/dgx/dgx-station-user-guide/index.html
[4] https://docs.nvidia.com/dgx/dgx-os-6-user-guide/upgrading.html
[5] https://nvidia.custhelp.com/app/answers/detail/a_id/5449/~/security-bulletin:-nvidia-dgx-2,-dgx-station-a100,-and-dgx-a100---march-2023
[6] https://www.robusthpc.com/wp-content/uploads/2021/11/nvidia-dgx-station-a100-system-architecture-white-paper_published.pdf
[7] https://docs.nvidia.com/dgx/dgx-os-5-user-guide/upgrading.html
[8] https://www.rgb.com/sites/default/files/legacy-products/firmware-updates/DGxfirmwareinstructions.pdf