-
Notifications
You must be signed in to change notification settings - Fork 520
Description
/!\ To report a security issue please follow this procedure:
[https://github.com/OpenNebula/one/wiki/Vulnerability-Management-Process]
Description
In Sunstone when working with PCI devices, the logic for getting the selected device is based on a DEVICE;VENDOR;CLASS comparison directly on the host monitoring data. This creates a problem for vGPU devices, as they all share the same device, vendor and class. Instead the selection should be based on the SHORT_ADDRESS of the device, and this should also be what gets stored on the actual template itself.
In its current state, if more than 1 vGPU device is present, its not possible to accurately select the correct device and fetch its gpu profiles, as the current logic always returns the first found vgpu device in the monitoring array.
To Reproduce
Create a host with a set of 1 or more vgpu devices and attempt to select a gpu profile from one of the latter vgpu devices in the monitoring array. The list will either be empty or display the wrong set of profiles, depending on the first vgpu devices monitoring data.
Expected behavior
The device selection should be based on SHORT_ADDRESS to correctly identify it.
Details
- Affected Component: [Sunstone]
- Hypervisor: [N/A]
- Version: [7.0.2]
Additional context
Add any other context about the problem here.
Progress Status
- Code committed
- Testing - QA
- Documentation (Release notes - resolved issues, compatibility, known issues)