GPU tab shows "No data available"

GPU tab shows "No data available"

GPU Stats Enabled but No GPU Data Collected (Linux)

Problem

GPU Stats is enabled for the Linux server, but no GPU metrics are displayed because the collection returned no data. 

Cause

Applications Manager collects GPU metrics using the nvidia-smi utility. If the command does not return any GPU information.

Resolution

Perform the following checks to identify the root cause:

1. Verify NVIDIA GPU Availability

Ensure that the monitored server is equipped with an NVIDIA GPU.

Note: GPU monitoring currently supports NVIDIA GPUs only. AMD GPUs are not detected or monitored.

2. Verify nvidia-smi Accessibility

Confirm that:

  • NVIDIA drivers are installed correctly.

  • The nvidia-smi utility is installed.

  • nvidia-smi is available in the PATH of the SSH monitoring user, not just the root user.

You can verify this by logging in as the monitoring user and executing:

which nvidia-smi
nvidia-smi

3. Validate the GPU Collection Command

Log in to the server using the same SSH user configured in Applications Manager and execute the following command:

Quoteecho -e "===NVIDIA_GPU===\n$(LANG=C nvidia-smi --query-gpu=index,uuid,name,driver_version,utilization.gpu,utilization.memory,memory.total,memory.used,memory.free,temperature.gpu,power.draw,power.limit,fan.speed,clocks.sm,clocks.mem,pstate,clocks.max.sm,clocks.max.mem,compute_mode,display_active,persistence_mode --format=csv,noheader,nounits 2>/dev/null)"

Expected Result

The command should return one or more CSV rows similar to:

===NVIDIA_GPU===
0, GPU-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx, NVIDIA RTX A4000, ...

If No Output Is Returned

If the command produces only:

===NVIDIA_GPU===

or no GPU data at all, the issue originates from the target server rather than Applications Manager. Possible causes include:

  • NVIDIA driver is not installed or is corrupted.

  • NVIDIA driver service is not running.

  • GPU is not detected by the operating system.

  • The SSH monitoring user does not have permission to access the GPU.

  • nvidia-smi cannot communicate with the NVIDIA driver.

Resolve the server-side issue first, then verify that the command returns valid GPU information.

4. Verify GPU Ignore Configuration

Check whether the GPU model is configured to be ignored.

Navigate to:

Settings → Performance Polling → Server → Linux → GPU Stats → Ignore Specified Components

If the GPU name appears in the GPUStats_Ignore list, Applications Manager intentionally skips collecting metrics for that GPU. Remove the GPU from the ignore list if monitoring is required.

Summary

If GPU Stats is enabled but no data is displayed:

  1. Verify that the server contains an NVIDIA GPU.

  2. Confirm nvidia-smi is installed and accessible to the SSH monitoring user.

  3. Execute the GPU collection command manually and verify that it returns CSV data.

  4. If the command returns no GPU information, resolve the underlying server or driver issue.

  5. Ensure the GPU is not configured in the GPUStats_Ignore list.