I’ve tried Apacer AS2280Q4 2TB and ADATA SWORDFISH 500 GB. Both report
nvme nvme0: globally duplicate IDs for nsid 1
since Linux 5.19, if I attach more than one.
Only the first drive is seen by the system. Workaround so far has been to stay on 5.15, but that’s not a viable long-term solution.
This error has been known for quite some time, and has been fixed downstream for specific distros and ssd models. Is there any chance the manufacturers will start to assign unique ID’s to each drive, or mainline implements usable a universal workaround?
I’ve never heard of a drive not working on linux.
Maybe its a platform limitation?
What system are you running and what motherboard?
MSI B450 MORTAR MAX, and ASROCK B550 Phantom Gaming
Proxmox 7.4
Motherboard doesn’t really matter though. All nvme ssds are detected by linux 5.15. From 5.19 up, only the first one is.
I’ve had issues with that particular kernel on one of my servers. It might be best to hold back the kernel until these bugs are fixed.
Its a defect in the device that can be fixed by the vendor, worked around by kernel devs, or by end user if they are willing to build a kernel
Your nvme drives don’t support providing proper device UUIDs instead it results in “0” which of course is the same on both. The actual fix is for your vendor to get their shit together which might mean updating to a new firmware version IF there is a fix. The next best thing would be for the kernel devs to specifically work around your broken model. Such workarounds will appear in new versions of kernels and may not be ported to older versions. See this thread on kernel.org https://bugzilla.kernel.org/show_bug.cgi?id=216049
If you are feeling very adventurous its possible to do this fix yourself in a kernel you build. It mostly entails figuring out how to build and install the kernel and pasting some text not learning c
https://unix.stackexchange.com/questions/711739/globally-duplicate-ids-for-nsid
