Sometimes crazy changes need to be performed to adjust how storage is laid out and arranged. In this instance converting an existing (but non-critical) RAID1 mdadm array to RAID0 (even less critical) is the far-fetched requirement.
Start by checking the status of the existing array and note down the necessary identifiers:
[root@srv /]# mdadm --detail /dev/md1 /dev/md1: Version : 1.0 Creation Time : Tue Aug 21 17:27:41 2012 Raid Level : raid1 Array Size : 524287860 (500.00 GiB 536.87 GB) Used Dev Size : 524287860 (500.00 GiB 536.87 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Intent Bitmap : Internal Update Time : Sat Feb 18 14:43:21 2023 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Consistency Policy : bitmap Number Major Minor RaidDevice State 3 8 17 0 active sync /dev/sdb1 2 8 1 1 active sync /dev/sda1
Ensure that any filesystem residing on the array is unmounted (hence why I called it non-critical before).
Then run the necessary command:
[root@srv /]# mdadm /dev/md1 --grow --level=0 mdadm: level of /dev/md1 changed to raid0
This will cause mdadm remove one of the members from the array and basically turn it into a single-disk RAID-ready array:
Number Major Minor RaidDevice State 2 8 1 0 active sync /dev/sda1
Proceed to add back the dumped/removed/missing member:
[root@srv /]# mdadm /dev/md1 --grow --raid-devices=2 --add /dev/sdb1 mdadm: level of /dev/md1 changed to raid4 mdadm: added /dev/sdb1 mdadm: Need to backup 8K of critical section..
The message says that array is being converted to RAID4 – that will only be temporary while the array is rebuilt and will change to 0 when completed.
[root@srv /]# cat /proc/mdstat Personalities : [raid0] [raid1] [raid6] [raid5] [raid4] md1 : active raid4 sdb1[3] sda1[2] 524287860 blocks super 1.0 level 4, 4k chunk, algorithm 5 [3/2] [U__] [=>...................] reshape = 0.1% (1046620/524287860) finish=208.2min speed=41864K/sec
In case you get a mdadm: failed to update superblock during re-add error at this step, try zero-erasing the problematic partition to remove its (previous) RAID signature:
[root@srv /]# dd if=/dev/zero of=/dev/sdb1 bs=10M status=progress 536745082880 bytes (537 GB, 500 GiB) copied, 3527 s, 152 MB/s dd: error writing '/dev/sdb1': No space left on device 51201+0 records in 51200+0 records out 536870912000 bytes (537 GB, 500 GiB) copied, 3565.52 s, 151 MB/s
Then retry the add.
When the reshape is done, check the status:
[root@srv ~]# mdadm --detail /dev/md1 /dev/md1: Version : 1.2 Creation Time : Sat Feb 18 19:22:47 2023 Raid Level : raid0 Array Size : 1048311808 (999.75 GiB 1073.47 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Sat Feb 18 19:22:47 2023 State : clean Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Layout : -unknown- Chunk Size : 512K Consistency Policy : none Number Major Minor RaidDevice State 0 8 1 0 active sync /dev/sda1 1 8 17 1 active sync /dev/sdb1
