Hi, On Wed, 2021-02-10 at 19:59:24 +0100, Thomas Lange wrote: > I think we want to match all real disks which can be identified by a > name, model, serial number. Maybe also virtualized disks which have > such an entry at by-id.
I agree with that. Hardware modifications will result in by-path entries changing (I've got still that 20-year old ZFS demonstration in mind where USB keydrives were pulled and rearranged in random patterns)... On Thu, 2021-02-11 at 09:53:02 +0100, Thomas Lange wrote: > [~]$ udevadm info -r --query=symlink /dev/sda > /dev/disk/by-id/ata-WDC_WD3000FYYZ-01UL1B0_WD-WMC1F0499068 > /dev/disk/by-path/pci-0000:01:00.0-sas-exp0x500304800096b4bf-phy17-lun-0 > /dev/disk/by-id/wwn-0x50014ee35570a5e8 I'm also seeing this, and while I prefer unique IDs like the WWN this cannot be found everywhere. What about disk=`udevadm info -r --query=symlink /dev/sda | xargs -n1 | sort | head -n1` which will select a by-id entry if that exists, and fall back to "something else" in other cases? The sorting will also prefer ata and scsi entries over wwn ones though... - S
