Troubleshooting Guide¶
This guide covers common issues you may encounter when using Intel® ESQ and provides solutions for each.
1. Idle memory retention after test execution¶
After test execution, idle memory usage may increase due to accumulated slab and filesystem cache. This can leave insufficient memory for subsequent workloads.
Solution: Force a filesystem cache cleanup by dropping dentries, inodes, and page cache:
2. Docker* permission denied¶
When running containerized tests, you may see an error such as permission denied while trying to connect to the Docker daemon socket.
Solution: Add your user to the docker group and refresh your session:
Verify access by running:
If the issue persists, log out and log back in or reboot the system.
3. Test shows as skipped¶
Tests may appear as skipped for one of the following reasons:
- A cached result already exists for the same parameters.
- The system does not meet the test's hardware or software requirements.
- Required devices (GPU, NPU) are not available.
Solution:
-
Run without cache to force a fresh execution:
-
Check system requirements:
-
Review debug output for device availability messages:
4. GPU not detected or inaccessible¶
Tests that require GPU access may fail with errors like No available devices found for configured device category or Required GPU hardware not available.
Possible causes:
- Intel GPU drivers are not installed.
- The GPU hardware is not present on the system.
- The user lacks permissions to access
/dev/drior/dev/accel.
Solution:
-
Verify GPU hardware is present:
-
Install or update Intel GPU drivers for your platform.
-
Confirm device access:
-
Run
esq infoto verify device detection.
5. NPU not detected¶
NPU-dependent tests may fail if the NPU driver is missing or not loaded.
Solution:
-
Install or update Intel NPU drivers.
-
Verify the NPU is detected:
-
Confirm the driver in use is
intel_vpu.
6. Profile not found¶
Using a shortened or incorrect profile name (for example, ai_vision) causes a profile-not-found error.
Solution: Always use the full dotted profile name. Run esq list to see available profiles:
Profile names follow the format profile.suite.ai.vision, profile.suite.ai.gen, and so on.
7. Test fails with no clear error¶
When a test fails but the CLI output does not show a clear root cause:
-
Run the test with debug-level logging:
-
Check the log files:
-
Review the Allure* report for detailed attachments and execution timeline:
-
Verify that asset preparation completed without errors.
-
For containerized tests, check Docker* container logs.
8. Stale cache causing unexpected results¶
Cached results from previous runs may cause tests to return outdated data, especially after code or configuration changes.
Solution:
-
Clean the cache:
-
Clean all data, including results and reports:
-
Run a specific test without cache:
9. RAPL power monitoring not working¶
Platform power monitoring through RAPL (Running Average Power Limit) may fail if device permissions are not configured.
Solution: Run the system setup script:
sudo bash -c "$(wget -qLO - https://raw.githubusercontent.com/open-edge-platform/edge-system-qualification/refs/heads/main/scripts/system-setup.sh)"
Note
RAPL is not supported on all platforms. Check your hardware documentation for compatibility.
10. Memory DIMM information unavailable¶
The esq info command may fail to retrieve memory DIMM details if dmidecode lacks the required capabilities.
Solution: Install the necessary packages and set the required capability with system setup script:
sudo bash -c "$(wget -qLO - https://raw.githubusercontent.com/open-edge-platform/edge-system-qualification/refs/heads/main/scripts/system-setup.sh)"