Put SD into board and wait for the setup to complete
login and run fdisk -l
Error is shown for the boot partition
Expected behaviour
No errors with the GPT partitions.
Actual behaviour
The boot partition is corrupt, but the backup is OK.
Disk /dev/mtdblock0: 16 MiB, 16777216 bytes, 32768 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 7B7A6FDE-8147-4712-A91F-524B1DDF8D33
The primary GPT table is corrupt, but the backup appears OK, so that will be used.
Extra details
Everything works fine, no issues while using the system whatsoever.
I’m building a RAID 1 with the board so that’s why I found this error. First I thought it was the cheap SD card I was using, so I bought a brand new SanDisk, but the error remains.
I flashed the SD cards multiple times, both of them, but the error is consistent, every single time I flash the image to the card and boot them.
Yeah, it doesn’t look like a DietPi issue, but actually something withmdadm.
After I’m done creating the array, I see this error on both devices in it.
cbaldan@R5BP:~$ sudo fdisk -l /dev/nvme0n1 /dev/nvme1n1
The primary GPT table is corrupt, but the backup appears OK, so that will be used.
Disk /dev/nvme0n1: 476.94 GiB, 512110190592 bytes, 1000215216 sectors
Disk model: WDC PC SN520 SDAPNUW-512G-1006
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 9BD20D8C-B6A5-4A75-A853-79003834B9CE
Device Start End Sectors Size Type
/dev/nvme0n1p1 2048 1000214527 1000212480 476.9G Linux RAID
The primary GPT table is corrupt, but the backup appears OK, so that will be used.
Disk /dev/nvme1n1: 476.94 GiB, 512110190592 bytes, 1000215216 sectors
Disk model: WDC PC SN520 SDAPNUW-512G-1014
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 8BA87062-B713-4498-9E5C-B5745EE0789B
Device Start End Sectors Size Type
/dev/nvme1n1p1 2048 1000214527 1000212480 476.9G Linux RAID
Isn’t it expected that adding drives to a RAID erases the partitions? GPT does have a backup table, but restoring it would then break the RAID, overwriting sections on the drive which are used for RAID metadata, isn’t it?
Though weird that fdisk even detects a GPT partitioned disk, even that the primary table is broken. Did you properly format the disks before adding them to the RAID?
E.g. our article on this topic states that existing partitions need to be removed: Using a RAID system with DietPi – part 1 – DietPi Blog
Overwriting the first MiB of the drive with zeros + partprobe should do the same, or erase traces of the prior partition table even better.
Thanks for the tutorial, I hadn’t seen Dietpi’s when I was looking for help.
I’ve just re-did the entire RAID setup: unmounted the device, deleted the RAID, cleared the superblock of both drives and deleted the partitions they had. The corrupt GPT table message was gone for both devices, and dietpi-drive_manager showed the drives exactly like in the tutorial - format required message.
Then I followed the tutorial to a T, the only difference to the other tutorial I followed was creating a partition of type fd on broth drives before the creating the array (https://youtu.be/CJ0ed38N8-s).
But still, right after I run the mdadm –-create command, the corrupt GPT table error is back on both drives.
I guess it’s just how it is, maybe the utils aren’t fully updated to support NVME drives yet??
cbaldan@R5BP:~$ sudo mdadm --create /dev/md1 --level=1 --raid-devices=2 /dev/nvme0n1 /dev/nvme1n1
(...)
cbaldan@R5BP:~$ sudo fdisk -l /dev/nvme1n1 /dev/nvme0n1
The primary GPT table is corrupt, but the backup appears OK, so that will be used.
Disk /dev/nvme1n1: 476.94 GiB, 512110190592 bytes, 1000215216 sectors
Disk model: WDC PC SN520 SDAPNUW-512G-1014
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 8BA87062-B713-4498-9E5C-B5745EE0789B
The primary GPT table is corrupt, but the backup appears OK, so that will be used.
Disk /dev/nvme0n1: 476.94 GiB, 512110190592 bytes, 1000215216 sectors
Disk model: WDC PC SN520 SDAPNUW-512G-1006
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 9BD20D8C-B6A5-4A75-A853-79003834B9CE
@Jappe thanks for the tip, I’m going to try btrfs since it’s considered mature for RAID1, but it’s odd that a tool that is almost 20 years old is still not considered 100% mature, with very divided opnions everywhere.
But I like the simplicity and flexibility it has over mdadm. I’m new to all this, I just hope I don’t lose all my pictures I’m self hosting in this array ¯_(ツ)_/¯ - immich.
I have searched the existing open and closed issues