Docs

Elastic Compute Service

GPU instances issue

Solution for RDMA network configurator exceptions

Solution for RDMA network configurator exceptions

This topic describes how to troubleshoot RDMA network configurator exceptions.

Procedure

Log in to the target instance. For operation details, refer to Logging in to Linux instances.

Confirm whether the service status of rdma-agent.service is active (exited). For more information about RDMA Agent, refer to Installing RDMA Agent in HPC-optimized instances.

systemctl status rdma-agent.service

A sample of normal output is as follows:

Perform the following operations when the service status is not active (exited):

Check whether your instance supports RDMA. The rdma-agent.service service is required only when the instance supports RDMA.

tip

Only HPC-optimized instances support RDMA. For more information, see HPC-optimized instances.

If the instance does not support RDMA network, it is considered normal for the rdma-agent.service service to have an abnormal status or not exist.

If your instance supports RDMA network, execute the following command to restart the service.

systemctl restart rdma-agent.service

After the service is restarted, execute the following command again to check whether the service status is active (exited).

systemctl status rdma-agent.service

To obtain technical support when the method described in this topic does not resolve your issue, submit a ticket or contact your account manager.

Last updated: 2026.06.18 17:49:51