ERROR, FAIL, and WARNING Status
Checks listed at the top of the list have higher priority and may have dependent checks further down the list. When diagnosing problems, start by troubleshooting the checks at the top of the list.
The output shows these items:
Check Name |
Name of the check that ran. |
Description |
A brief description of what information the check is reporting. |
Result status |
PASS - the check found the expected result. |
Examples of Result Status
[+] PASS - If the check was to find information such as a version number, a passing check will display that information.
[-] FAIL - If the check was to run a GPU check on a system that does not have a GPU, the check will fail and the description will indicate that the GPU is not present.
/!\ WARNING - If the check successfully runs, but can’t parse the incompatible or incomplete result, there will be a warning with a brief description as to why the information is incomplete or incompatible.
-
- (-) ERROR - check was not able to run. Possible causes:
-
current user does not have permissions to access information that the check is looking for such as a driver version, but the driver is not accessible to the current user.
software or hardware is not initialized.
Updating the Checks Database
Before troubleshooting a check, first ensure that the checks database is up to date with the latest version. Updating to the latest version will provide updates to output messages that may give you more detail about why a check has a status of ERROR or FAIL.
To update the database, run the --update argument:
python3 diagnostics.py --updateIf your machine does not have access to the internet for database updates, or if for some reason the database file has been deleted from the package (eg manual deletion, etc) it is possible to manually download the database file.
In this case, download the database and the --metadata.json file from the public github repository and put them in the {onediag_root}/databases folder.
Resolving Failed Checks
A status of FAIL indicates that the check ran successfully but the result of the check does not meet the expected requirements. The table below shows the expected result and how to fix the problem.
For example, running python3 diagnostics.py --select user_group_check intel_gpu_detector_check may produce this output:
Checks results:
======================================================================================================================
Check name: intel_gpu_detector_check
Description: This check shows which Intel GPU(s) is on the system based on lspci information and internal table.
Result status: PASS
======================================================================================================================
======================================================================================================================
Check name: user_group_check
Description: This check verifies that the current user is in the same group as the GPU(s).
Result status: FAIL
Current user is not part of the video group, contact the system administrator to add current user to the video group.
======================================================================================================================
2 CHECKS, 1 PASS, 1 FAIL, 0 WARNINGS, 0 ERRORS
Console output file: /home/test/intel/diagnostics/logs/diagnostics_select_intel_gpu_detector_check_hostname_20211123_103132327319.txt
JSON output file: /home/test/intel/diagnostics/logs/diagnostics_select_intel_gpu_detector_check_hostname_20211123_103132327361.json
The report was generated for the machine: hostname
by the Diagnostics Utility for oneAPI 2024.2.0
Result status: FAIL
Current user is not part of the video group, to add a user: sudo usermod -a -G video test.In this case, the output gives a possible solution to fix the failed check.
The table below shows likely solutions for a failed status for each available check:
Check that Failed |
Possible Solution |
|---|---|
gpu_backend_check |
OpenCL or LevelZero Driver is not loaded. Load the driver. Current user may not have permissions to access driver folder. Try running as sudo |
oneapi_toolkit_check |
Information only. If an Intel® oneAPI toolkit is found, the check will INFO. If an Intel® oneAPI toolkit is not found, the check will FAIL. |
driver_compatibility_check |
Check drivers to ensure they are compatible with installed oneAPI components. |
compiler_check |
Compiler is not installed. |
intel_gpu_detector_check |
Information only. If a gpu is found, the check will PASS. If a gpu is not found, the check will FAIL. |
base_system_check |
Information only. If a system information is found, the check will PASS. If system information is not found, the check will ERROR. |
hangcheck_check |
Hangcheck is currently enabled. Disable hangcheck to get this check to PASS. |
user_group_check |
Current user is not a part of the same group as the GPU. Add current user to the video group with the command sudo usermod -a -G video test . |
oneapi_env_check |
|
kernel_options_check |
Information only. If a system information is found, the check will PASS. If system information is not found, the check will ERROR. |
gpu_metrics_check |
|
oneapi_gpu_check |
Check was not able to run workloads on the GPU. Try running as sudo. If sudo does not work, GPU is not ready to run workloads. |
user_resources_limits_check |
When running a group of checks, you may have multiple failures simultaneously. Fixing one failure could potentially fix other failures, depending on their relationship.