Skip to content

add support for overriding fields in fabricmanager.cfg - #893

Open
tariq1890 wants to merge 1 commit into
mainfrom
fabric-mode
Open

add support for overriding fields in fabricmanager.cfg#893
tariq1890 wants to merge 1 commit into
mainfrom
fabric-mode

Conversation

@tariq1890

@tariq1890 tariq1890 commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

Allow gpu-driver-container users to override fabric manager config before starting the fabricmanager daemon on their NVSwitch GPU systems. With this change, the gpu-driver-container will the parse environment variables with the NVFM_CONFIG_ prefix and modify the /usr/share/nvidia/nvswitch/fabricmanager.cfg file with the desired fabric manager config before starting up fabricmanager.

@varunrsekar

Copy link
Copy Markdown

@tariq1890 In the interest of just tweaking fabric-mode, this PR looks correct to me.

But what about other configuration tweaks like the resiliency mode and the FM service connection method? Are they going to be handled in separate PRs?

@tariq1890 tariq1890 changed the title add support to set FABRIC_MODE in the fabricmanager.cfg file add support for overriding fields in fabricmanager.cfg Aug 3, 2026
Signed-off-by: Tariq Ibrahim <tibrahim@nvidia.com>
@tariq1890

Copy link
Copy Markdown
Contributor Author

Thanks for taking a look @varunrsekar. I have extended the implementation to make it more generic. Now, users can override the fabric manager config field of their choice by providing the environment variables with this format - NVFM_CONFIG_<FABRIC_MANAGER_CONFIG_FIELD>=<FABRIC_MANAGER_CONFIG_VALUE>

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants