I am getting a Duplicate UUID error when I attempt to install Fedora 37/38 to my hard drive which contains a linux raid1 array.
This is the message I’m getting:
Duplicate UUID ‘97d392c8-01’ found for devices: ‘sdb1’ and ‘sda1’.
This is usually caused by cloning the device image resulting in duplication of the UUID value which should be unique. In that case you can either disconnect one of the devices or reformat it.
This UUID looks to be the Disk Identifier of the /dev/sda and /dev/sdb disks. Here is the output of fdisk -l:
Disk /dev/sda: 1.82 TiB, 2000398934016 bytes, 3907029168 sectors
Disk model: ST2000DM006-2DM1
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disklabel type: dos
Disk identifier: 0x97d392c8
Device Boot Start End Sectors Size Id Type
/dev/sda1 2048 3905970175 3905968128 1.8T fd Linux raid autodetect
/dev/sda2 3905970176 3907028991 1058816 517M fd Linux raid autodetect
Disk /dev/sdb: 1.82 TiB, 2000398934016 bytes, 3907029168 sectors
Disk model: ST2000DM006-2DM1
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disklabel type: dos
Disk identifier: 0x97d392c8
Device Boot Start End Sectors Size Id Type
/dev/sdb1 2048 3905970175 3905968128 1.8T fd Linux raid autodetect
/dev/sdb2 3905970176 3907028991 1058816 517M fd Linux raid autodetect
Can you advise as how to rectify this install issue?
This issue seems to be a result of the installer not seeing the devices as raid1 but rather as different devices.
What is the output of cat /proc/mdstat?
Also, the drive are formatted with MSDOS partition table and not GPT which may be a factor.
When posting text from your screen please use the Preformatted text tags available with the </> button on the toolbar above so it retains the formatting seen on your screen. I am editing your post above to do just that for you.
Looking at the fdisk output again I am seeing only 2 partitions. One is 1.8 TB and one is 500 MB. This seems like you have already partitioned the devices so there would be no space for an install.
If you were to install in legacy mode with ext4 I could see that fedora may install with only 2 partitions. A ‘bios boot’ partition and the root ext4 partition, but it would require custom partitioning to achieve that.
What is already on those drives? If nothing then it would be relatively simple to wipe out the raid array by totally removing the partitioning then recreating the raid array while doing the install.
If there is data that you wish to preserve then restructuring the array first would be the ideal way to go. It may even mean failing one of the drives out of the array so it can be disconnected, then manipulating that drive before adding it back so it gets a proper UUID.
The 500MB raid array just contains /boot. The 1.8TB raid array contains /, /home, /var, /swap logical volumes. Essentially all the data I want to preserve.
I don’t seem to have the option to assemble the arrays when I select Install Fedora. The Install to Hard Drive Window appears with the Welcome to Fedora window - What Language would you like to use during the installation process. Then a popup window appears with this text:
There is a problem with your existing storage configuration or your initial settings, for example a kickstart file. You must resolve this matter before the installation can proceed.
Duplicate UUID '97d392c8-01' found for devices: 'sdb1' and 'sda1'.
This is usually caused by cloning the device image resulting in duplication of the UUID value which should be unique. In that case you can either disconnect one of the devices or reformat it.
I agree 100%.
The raid array is the device, not the raw disk.
It also seems odd that you would have /boot in an LV. This means grub must load the LVM modules before the system can boot. This certainly is possible but adds one more layer of complexity to the boot process.
The only data that you probably need to be concerned with when doing a new install is /home. Everything else should be wiped out and built new. Your swap partition is likely not required unless you intend to suspend/hibernate and the space would be needed for that.
If you are using VMs then saving those images may be a good thing as well, but will likely require some manipulation to save them without extra drive space.
More info can be gotten with vgscan, vgdetail, lvscan, & lvdetail
The reason I made boot in an LV is that if one of the disks in the raid failed I would still be able to boot the system since boot records are on both disks. The same applies for swap.
I have data in /var in addition to /home which I’d need to maintain.
I have all the data backed up so I could start from scratch with a new install though it would take time to recover the data I have in backup.
Backup the exact data from /var that you need to keep then you would not even need the /var partition. The data could be restored after the new install into /var as part of the root partition.
A new install will require formatting of /, /boot, /var if /var is retained as a separate partition.
A backup of /home would be a good idea but is often not necessary if one does a custom install and tells the system to not format /home. After all, you already have the partitions defined, and for most only /home contains data that must be retained.
Having /var as a separate partition is mostly a legacy from times when drives were small and the dynamic nature of /var tended to fill the root file system and cause crashes. Keeping that data in a separate partition tended to prevent such crashes. Usually that is not really an issue for most any more, but do as you choose.
Swap in fedora is now zram by default so the physical swap is not normally necessary except for special cases (very low RAM, suspend, etc.) Actually having swap in raid tends to slow things down since writing to swap is duplicated in time requirements.
The sda1/sdb1 devices are 1M bios boot partitions which are required by the fedora live installer. The md126 (sda2/sdb2) raid array has two logical volumes, / and /home. I no longer have a /var LV based on your point about it being legacy and no longer required. I no longer have a partition for swap either.
The md127 (sda3/sdb3) raid array is for the boot partition. sda3/sdb3 aren’t LVM members, LVM isn’t allowed for boot partitions with the Fedora Live 38 installer.
I guess the next step is to fail one of the md126/md127 members and test that the system will still boot.