Docs

GPU Compute Service

Monitoring

RDMA network monitoring (hpcpni2)

After creating an HPC-optimized hpcpni2 instance, you can manually install the plugin for RDMA network observability enhancement in the instance. This plugin is used to monitor the health status of the RDMA network.

Scenario

This feature is applicable for an HPC-optimized hpcpni2 instance running on a public image with "RDMA network observability enhanced" disabled. For more "RDMA network observability enhanced" information, refer to Creating HPC-optimized hpcpni2 instances.

It is also applicable for an HPC-optimized hpcpni2 instance running on a custom image.

Install RDMA network enhancement plugin

In the left-side navigation pane, select "Instances & Images > Instances".

In the top menu bar, select the region of the target instance.

On the Instances page, use the root account to log in to the Linux instance.

After successful login, execute the corresponding commands below to complete the installation.

China (Beijing):

bash -c "$(wget -q -O - wget https://rdma-tool-beijing.tos-cn-beijing.volces.com/install_rdma_monitor.sh)"

China (Shanghai):

bash -c "$(wget -q -O - wget https://rdma-tool-shanghai.tos-cn-shanghai.volces.com/install_rdma_monitor.sh)"

China (Guangzhou):

bash -c "$(wget -q -O - wget https://rdma-tool-guangzhou.tos-cn-guangzhou.volces.com/install_rdma_monitor.sh)"

View installation result.

If the response is as follows, the plugin is installed. After three to five minutes, the plugin can collect monitoring data as expected.

rdma-monitor installed
rdma-monitor.service start

Uninstall RDMA network enhancement plugin

If you no longer need the feature of RDMA network observability enhancement, you can uninstall the RDMA network enhancement plugin.

Execute the following command to complete the uninstallation.

bash /usr/local/rdma-monitor/uninstall.sh

Last updated: 2025.12.03 16:28:04