Software RAID with mdadm

Linux software RAID uses mdadm to group multiple disks into a fault-tolerant volume. This guide applies to HolyCloud dedicated servers where you manage disks (RAID 1 mirror or RAID 5 with parity).

Prerequisites

• Dedicated server with at least two identical disks (RAID 1) or three+ (RAID 5)
• IPMI/KVM access if system no longer boots
• Full backup before any partition manipulation
• Debian/Ubuntu distribution or manual installation

Warning: creating RAID on disks containing data erases them. Check lsblk and labels.

Disk inventory

lsblk -o NAME,SIZE,TYPE,MODEL,SERIAL
cat /proc/mdstat
mdadm --detail --scan

Install mdadm if missing:

apt update && apt install -y mdadm

Create RAID 1 (mirror)

Example: /dev/sdb and /dev/sdc, partition type Linux RAID (fd for MBR, raid for GPT).

parted /dev/sdb --script mklabel gpt mkpart primary 1MiB 100% set 1 raid on
parted /dev/sdc --script mklabel gpt mkpart primary 1MiB 100% set 1 raid on

Create array:

mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sdb1 /dev/sdc1

Follow prompts (optional alert email). Format:

mkfs.ext4 /dev/md0
mkdir -p /mnt/raid
mount /dev/md0 /mnt/raid

Persistence: mdadm.conf and fstab

Capture configuration:

mdadm --detail --scan | sudo tee -a /etc/mdadm/mdadm.conf
update-initramfs -u

UUID for fstab:

blkid /dev/md0

Add to /etc/fstab:

UUID=xxxxxxxx  /mnt/raid  ext4  defaults,nofail  0  2

Create RAID 5 (overview)

Three disks minimum:

mdadm --create /dev/md1 --level=5 --raid-devices=3 /dev/sdd1 /dev/sde1 /dev/sdf1

Usable capacity ≈ (N-1) × smallest disk size.

Monitoring

Real-time state:

cat /proc/mdstat
mdadm --detail /dev/md0
watch -n 2 cat /proc/mdstat

Alert email: configure MAILADDR in /etc/mdadm/mdadm.conf and a local MTA.

Replace a failed disk

Mark disk failed (example sdb1):

mdadm /dev/md0 --fail /dev/sdb1
mdadm /dev/md0 --remove /dev/sdb1

After physical replacement, recreate identical partition, then:

mdadm /dev/md0 --add /dev/sdb1

Resync appears in /proc/mdstat (recovery).

Degraded at boot

mdadm --assemble --scan
mdadm --detail /dev/md0

Troubleshooting

| Situation | Command |

|-----------|----------|

| RAID stopped | mdadm --assemble /dev/md0 /dev/sdb1 /dev/sdc1 |

| Inconsistent superblock | mdadm --examine /dev/sdb1 (compare members) |

| Force assemble (danger) | Last resort — HolyCloud support |

Need help?

For hardware RAID (LSI/HPE controller), specify the model in your ticket — procedure differs from mdadm.