NVIDIA BCM Scope Of Support
C.1 Included In BCM Support
NVIDIA Base Command Manager (BCM) technical support has a standard scope of coverage that includes the following:
C.1.1 Support For Cluster Installation
The BCM support team will help install BCM on a supported Linux base distribution, using any of the documented BCM features.
If the add-on installation method is used (as opposed to the bare-metal installation method), then installation on top of a default minimal installation of the base Linux distribution is supported. A default minimal installation of the Linux base distribution means selecting a minimal set of packages during the installation and leaving all settings at their defaults.
C.1.2 Support For Cluster Configuration
The BCM support team can help create configurations to meet the specific needs of a customer using any of the documented BCM features.
If Security-Enhanced Linux (SELinux) is enabled on the head node or compute nodes and a problem arises, then a customer may be asked to demonstrate that the problem also exists when SELinux is disabled, in order to confirm that the problem is not caused by an SELinux policy configuration.
C.1.3 Support For Integration Of Cluster In Site Infrastructure
Site infrastructures vary greatly, and so support is not provided for the third-party hardware and software that may be involved. Nevertheless, the BCM support team will do its best to provide support on matters related to cluster integration within a site’s infrastructure.
Information on how to accomplish many integration tasks is provided in the BCM product manuals and in the Knowledge Articles of the NVIDIA Enterprise Support Portal, and is often all that is needed to get the job done. If not, then the BCM support team may be able to help. The BCM support team will ask that the customer demonstrate how to accomplish the aspect of integration on an ordinary machine running the same Linux distribution as the cluster. Once the BCM support team has that information, it can provide instructions on how to accomplish the same task using BCM.
C.1.4 Support For Integration With Microsoft Active Directory
NVIDIA provides documentation that describes how to integrate the BCM cluster with Active Directory; however, that integration requires help from the Windows Server administrators. It is the responsibility of the BCM administrator to communicate with the Windows Server administrators to get the required information for the chosen integration method.
Debugging Active Directory integration issues is beyond the scope of standard support.
C.1.5 Support For First-Party Hardware Only
The BCM support team does not provide support for third-party hardware issues. When it is unclear whether an issue is due to hardware or software problems, the BCM support team will work with a customer to determine the source of the problem.
C.1.6 Support For Software Upgrades
Supported customers receive access to all new software releases, and the BCM support team manages any issues that arise as a result of an upgrade.
To identify versions of its software, BCM uses pairs of numbers (e.g., 8.0 or 9.2). The first number denotes the major version of the software and the second number denotes the minor version.
An upgrade within the major version (e.g., from 9.1 to 9.2) is called a minor upgrade. An upgrade from one major version to the next (e.g., from 8.2 to 9.0) is called a major upgrade.
Minor upgrades and recent major upgrades can be done in place. The BCM support team will enable smooth transitions during software upgrades.
C.2 Excluded From BCM Support
Support coverage for BCM excludes the following:
C.2.1 Help With Third-Party Software
Third-party software is all software not developed by NVIDIA, even though it may be packaged or integrated with BCM. Examples include, but may not be limited to, the Linux kernel, all software belonging to the Linux base distribution (e.g., Red Hat), Open MPI, ScaleMP, and workload management systems such as Slurm, SGE, LSF, Torque, PBS Pro, and UGE.
An exception will be made if a customer demonstrates that NVIDIA packaged an application incorrectly, or if the customer demonstrates that the integration in the stock user environment is incorrect.
C.2.2 Help With User Applications
No support can be provided on issues related to compiling or running user applications, i.e., applications that are not packaged as part of BCM and which are installed onto a cluster by a customer or other individual.
If a customer suspects the issue relates to an NVIDIA product, then the problem must be demonstrated using a small application that will allow NVIDIA engineers to reproduce the problem.
C.2.3 Help With Workload Management
No support will be provided on issues relating to the workload management system . The only exception is during the initial installation and configuration and in cases where a customer demonstrates that an issue is caused by incorrect integration between BCM and the workload manager.
C.2.4 Help With Performance Issues
No support will be provided to trace performance issues in the cluster. The only exceptions are performance issues related to software components developed by NVIDIA.
C.2.5 Root-Cause Analysis
The NVIDIA support engineers will do their best to determine the cause of a failure, but cannot guarantee that all issues can be traced down to a root cause.
C.2.6 Phone Support
Phone support is not provided by the BCM support team. This is because support via a written ticket system is generally more efficient, as it usually allows better tracking, more precision in conveying meaning, easier information gathering, and a more optimal saving of information for later internal reference and search queries.
C.3 Support Outside Of The Standard Scope—Getting Professional Services
Support for items outside of the standard scope are considered professional services.
The BCM support team normally differentiates between
-
regular support (customer has a question or problem that requires an answer or resolution), and
-
professional services (customer asks for the team to do something or asks the team to provide some service).
Professional services outside of the standard scope of support can be purchased via the NVIDIA Enterprise Services page at: