Release notes for NVIDIA Base Command™ Manager (BCM) 11.32.0

Released: 12 March 2026

General

New Features

  • Support for GB300 and JFab in OCI

  • Support for GB300 in AWS

  • Support for BaseOS 7.4.0 for x86_64 and aarch64

  • Support for Slurm 25.11

  • Improved support in bcm-netautogen and bcm-pod-setup for streamlining GB300 and B300 deployments

  • Added CUDA 13.1 packages

  • Added cm-hpcx-doca-ofed-cuda12 and cm-hpcx-doca-ofed-cuda13 packages

  • Added support for RHEL add-on installations using the BCM Rocky ISO

  • Added a feature in cm-clone-install to allow specifying a custom disk layout for HA setup

  • Added a new package cm-config-apt-nmc that enables Debian repositories for installing decoupled NMC plugins

  • Updated Ubuntu 24.04 base OS to 24.04.3

  • Updated Slurm 25.05 to 25.05.6

  • Updated Topograph to v3.8.0

  • Updated containerd to 2.1.6

  • Updated cm-nvhpc to 25.11

  • Updated cm-nvidia-container-toolkit to v1.18.2

  • Updated cm-openssl to 3.5.5

  • Updated enroot to 4.0.1

  • Updated golang-go-latest to 1.25.6

  • Updated pyxis to 0.21.0

Known Issues

  • Following the update of the cm-openssl package, clusters using cluster-extension may fail to pivot to the local root after an update of the packages in the software image. Administrators should terminate and recreate the compute nodes after the update, or alternatively, add a finalize script for the affected nodes or categories that runs sed ‘s/^dh dh1024.pem/dh none/’ -i /localdisk/etc/openvpn/vpn.0.conf to ensure proper OpenVPN configuration and a successful pivot.

Changed

  • CMDaemon now always writes the DisableRootJobs parameter for all Slurm partitions in the slurm.conf configuration file. If DisableRootJobs has been manually set to Yes at the global level in slurm.conf, that configuration will no longer take effect. Administrators who need to disable root jobs must set the Disable Root property to Yes for the relevant Slurm partitions in BCM (for example, via cmsh -> wlm -> slurm -> jobqueues -> set disableroot yes or the equivalent path in Base View).

Fixed Issues

  • Fixed a crash in cmsh when adding a category without specifying a name

Removed Features

  • The Kubernetes Dashboard is no longer supported upstream, and can no longer be installed with cm-kubernetes-setup

COD-AWS

New Features

  • Added support for SSO authentication

CMDaemon

New Features

  • Improved Topograph integration: CMDaemon now stops issuing additional requests after Topograph returns a 50x response

  • Improved support in minimal configuration compute nodes to allow for offloading Slurm

  • CMDaemon will now generate a minimal valid topology configuration when it is unable to fetch the topology from Topograph

  • CMDaemon will no longer write the deprecated Slurm parameter AccountingStorageUser in slurm.conf

  • Added a command in cmsh to see active monitoring tasks: device taskrunnerdetails

Fixed Issues

  • A configuration issue inside the NVOS json ZTP file

  • An issue with monitoring that resulted in samplers with fuzzy-offset=1 to stop producing data after 1 iteration

  • Buffer overflow issue in the httpd handler

Base View

New Features

  • Added device status monitoring mode to Rack View in Base View

  • Highlight GB300 racks in Rack View in Base View with a gold color

  • Added Topograph settings to WLM wizard in Base View

cm-mission-control-setup

New Features

  • Added support for decoupled TUI wizards for DPS, NetQ, AHR, and AJR. The wizards are now delivered as separately installable packages but remain accessible through the existing cm-mission-control-setup interface

cm-kubernetes-setup

New Features

  • Added an option in cm-kubernetes-setup to install Kubernetes software in a software image without re-running a full Kubernetes setup

Deprecated or Removed Features

  • ingress-nginx is now a deprecated feature in cm-kubernetes-setup

  • Support for Grafana promtail installation has been removed from cm-kubernetes-setup

Fixed Issues

  • An issue with removing Kubernetes when kubecluster was not removed from KubeApiServerProxyRole

cm-wlm-setup

New Features

  • Improved default Slurm configuration settings for new Slurm installations to enhance the security of the setup, including updates to AccountingStorageType, ConstrainCores, ConstrainRAMSpace, and PrivateData

  • Fixed a rare enroot issue where the permissions of /var/lib/enroot could occasionally be set incorrectly

cmha

New Features

  • Added an option in cmha to skip replication of Slurm accounting database when the Slurm database is not on the head node