Node Validations
As a node group can contain multiple nodes, it is important to validate the nodes to ensure the node group is healthy.
DGX Cloud Lepton provides a simple way to run node validations for you to check the node group health, including the following validations:
- IB Ping
- Ethernet Ping
- NCCL Test
- GPU Burn
- FIO Test
- DCGM Diagnostics
Run a Validation
You can run a node validation and view the past validations for the node group under the Validations tab.
Click on the Run node validation dropdown, you can select the validation type you want to run.
If the validation has been run, there will have a indicator to show the latest validation result of this validation type: green for running, gray for completed, and red for failure.
View Validation Result and Logs
Click on the name filed of the validation table row, you will be navigated to the validation result page, where you can view the validation result and logs.
- The result tab will show the final validation result.
- The logs tab will show the logs of the validation, categorized by node. You can change the node you want to view logs by selecting from dropdown on the top right corner.
