Blivet does not show RAID partition device in the DeviceTree. This is the relevant part from the journal (attach whole journal as well) DEBUG:blivet: DeviceTree.get_device_by_name: name: SOMERAID1 ; incomplete: False ; hidden: False ; DEBUG:blivet: DeviceTree.get_device_by_name returned None DEBUG:blivet: PartitionDevicePopulator.run: name: SOMERAID1 ; DEBUG:blivet: DeviceTree.get_device_by_name: name: SOMERAID1 ; incomplete: False ; hidden: False ; DEBUG:blivet: DeviceTree.get_device_by_name returned None INFO:program:Running... udevadm settle --timeout=300 DEBUG:program:Return code: 0 DEBUG:testdata:sysfs attr /sys/class/block/sr0/device/model: 51454d55204456442d524f4d202020200a DEBUG:blivet: DeviceTree.get_device_by_name: name: SOMERAID ; incomplete: False ; hidden: False ; DEBUG:blivet:existing RAID raid0 size == 29.98 GiB DEBUG:blivet: DeviceTree.get_device_by_name returned existing 29.98 GiB mdarray SOMERAID (300) with existing disklabel DEBUG:blivet:existing RAID raid0 size == 29.98 GiB ERROR:blivet:ignoring partition SOMERAID1 on SOMERAID: parted disk not found DEBUG:blivet:no device obtained for SOMERAID1 Reproducible: Always Steps to Reproduce: 1. Boot 42 workstation installer ISO 2. have vda/vdb/vdc disks all empty 3. put biosboot and /boot on vda 4. create raid (level 1) in cockpit-storage (NAME SOMERAID) 5. delete the raid SOMERAID device 6. create raid (level 0) in cockpit-storage (NAME SOMERAID) 7. Put GPT table and create / partition on that Actual Results: Try to return to installation and see this error: 'You have not defined a root partition (/), which is required for installation of Fedora to continue.' This is because SOMERAID1 partition is ignored by blivet as seen in the journal above. Expected Results: SOMERAID1 should not be ignored. I have an automated test reproducer. I can help with extra debugging.
Created attachment 2083399 [details] journal.output
Sorry one small correction. This is only reproduceable if between steps 4 and 5 the user tries to exit cockpit-storage, that's trigerring the re-scan.
I still see this in our upstream CI on rawhide. Attaching journal log.
Created attachment 2087892 [details] journalctl output
The test that reproduces this on CI is this one: https://github.com/rhinstaller/anaconda-webui/blob/74868794683d4b876c8fdf336f173c3f2a477a64/test/check-storage-cockpit#L511