brandon3055

I'm not sure if this should be considering a bug report or a feature request so mods feel free to move it if it's in the wrong place. So long story short i recently upgraded my unraid system and in the process my pci configuration changed. Previously i had a network card in the top slot which was a PCIEx1. With my new config i have removed that network card and the top slot is now a PCIEx16 containing my LSI card. You can probably see where this is going... The other day i tried to start an old vm that i forgot i had previously passed the network card through to but now the pcie address that previously belonged to the network card belongs to the LSI card responsible for all the drives in my array. What happened next was many hours of of trying to figure out wtf just happened and then repairing my array. Fortunately i was able to avoid data loss but then i almost made the same mistake again. Because while i was initially troubleshooting why my array just imploded i discovered the LSI card was passed assigned to the vm and tried to unassign it. After i recovered the array i went back to the vm settings and the LSI card was no longer even a passthrough option so i assumed i had successfully unassigned it. But just to be safe i checked the XML and found that it was actually still assigned to the vm. I understand this is a very edge case issue that may not be solvable. But maybe unraid could keep a map of "pci address -> Device type" or something like that. Then give a warning when the device at an address assigned to a vm changes. Or even just do a sanity check each time you start a vm. Unraid obviously knows that LSI card should not be assigned to a vm so maybe when you go to start the vm it could stop you and be like "Hay this device probably shouldn't be assigned to a vm. Are you sure you want to continue?"