Description of problem: When deploying Swift following the best practise guide: https://access.redhat.com/articles/2745961 It created the entry in /etc/fstab using /dev/sdb instead of the UUID of the block device. This can lead to issues if the assignment changes. For example, a Kernel update has resulted in the internal card reader of a HP server being labelled as /dev/sdb instead of the previous /dev/sdc. Rebooting into the old kernel fixes is. It seems to only be an issue in fstab as swift.conf refers to /srv/node. So puppet-swift might not be the right place, please feel free to reassign the component as required. Version-Release number of selected component (if applicable): puppet-swift-12.4.0-4.el7ost.noarch How reproducible: Only if the kernel decides to mix the /dev/sdX devices around Steps to Reproduce: 1. Deploy as per best practise recommendations above 2. Update kernel 3. Hope it doesn't relabel the disks. Actual results: Failure during overcloud update as swift is unable to complete the update. Expected results: Device should be added to /etc/fstab by it's UUID as per Linux best practise. Additional info: $ cat etc/fstab LABEL=img-rootfs / xfs defaults 0 1 /dev/sdb /srv/node/sdb xfs noatime 0 0
Looks like a good idea, we should work on this for an upcoming release.
Proposed upstream patch: https://review.opendev.org/#/c/747444/
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Red Hat OpenStack Platform 16.1 bug fix and enhancement advisory), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHEA-2020:4284