MN_NVLINK_Compute_Tray_Failure : Compute Node Power Failure For now enabled EINJ using Compute Tray Power Off.
Compute Node Power Failure For now enabled EINJ using Compute Tray Power Off.
Bring down the Compute Node 1 while 2-N node try to access shared memory from Node 1 thorugh CUDA application .
Severity
Error
Alert Category
Power/Hardware Fault
User Action
- Check HMC event logs for specific power fault details.
- Perform an AC power cycle.
- If the issue persists, collect a diagnostic bundle using NVDebug.
- Contact Lenovo Support.
Give documentation feedback