Description of problem: There is a risk of a node having multiple disks and losing a disk then another disk will take it's device name. Ex. /dev/sda /dev/sdb /dev/sdc /dev/sdd Now /dev/sdc fails. The host reboots. What will happen is that /dev/sdd will now have the identifier /dev/sdc which will cause problems in our handling of the disks. Version-Release number of selected component (if applicable): How reproducible: always Steps to Reproduce: 1. Fail a disk. 2. Roboot the host 3. watch that the disks will be re-ordered Actual results: Expected results: We should refer to disks by their unique id that can be found at /dev/disk/by-uuid/<uuid> because this does not change on reboot. Additional info:
*** Bug 1337417 has been marked as a duplicate of this bug. ***
Jeff, I tried doing a patch and impact seems to be more and all the flows would be required to be verified as well. Also, it would be certainly risky at this moment to change something at framework level. So Nishanth suggested for 3.0..