Recently I upgraded my Ubuntu 11.04 server from 2x160GB drives to 2x500GB drives in a RAID1 configuration. While the physical disks show correct sizes in fdisk and sfdisk, the RAID array stubbornly remains at its original size:

# mdadm --examine /dev/sdb3
Used Dev Size : 152247936 (145.19 GiB 155.90 GB)
Array Size : 152247936 (145.19 GiB 155.90 GB)

The root cause lies in the RAID superblock metadata which retains the original device size. This is a common issue when using metadata version 0.90 (as shown by the "Magic: a92b4efc" in the examine output).

Current status shows:

# cat /proc/mdstat
md2 : active raid1 sdb3[0] sda3[1]
      152247936 blocks [2/2] [UU]

Here's how to properly resize the array and filesystem:

# First, stop the array (if possible)
mdadm --stop /dev/md2

# Reassemble with updated size detection
mdadm --assemble /dev/md2 /dev/sd[ab]3 --update=devicesize

# Verify the new size is detected
mdadm --examine /dev/sd[ab]3 | grep "Dev Size"

# Grow the array to maximum possible size
mdadm --grow /dev/md2 --size=max

# Check the new size in /proc/mdstat
cat /proc/mdstat

# Finally, resize the filesystem
resize2fs /dev/md2

If you encounter issues:

1. Ensure all partitions are properly aligned (check with parted)
2. Verify both disks show identical partition layouts
3. Consider updating to metadata version 1.2 for future flexibility

# To convert metadata version (be cautious!)
mdadm --grow /dev/md2 --metadata=1.2

When setting up new arrays, consider using:

mdadm --create /dev/md0 --level=1 --metadata=1.2 --raid-devices=2 /dev/sd[ab]1

This ensures the array will automatically recognize larger devices when replaced.

---

When upgrading RAID1 arrays with larger drives, many administrators encounter the same frustrating scenario:

# mdadm --grow /dev/mdX --size=max
# resize2fs /dev/mdX
# Nothing changes!

The root cause lies in how mdadm stores device size information in the metadata. When examining your array:

# mdadm --examine /dev/sdb3
Used Dev Size : 152247936 (145.19 GiB 155.90 GB)
Array Size : 152247936 (145.19 GiB 155.90 GB)

This metadata was created with the original 160GB drives and persists even after physical disk replacement.

Here's how to properly expand your RAID1 array:

First, verify current component sizes:

# cat /proc/partitions | grep sd[ab]3
   8        3  152248005 sda3
   8       19  152248005 sdb3

Now the critical step - reassemble with updated metadata:

# mdadm --stop /dev/md2
# mdadm --assemble /dev/md2 /dev/sda3 /dev/sdb3 --update=devicesize

Then grow the array:

# mdadm --grow /dev/md2 --size=max
# resize2fs /dev/md2

Check the new array size:

# mdadm --detail /dev/md2 | grep Size
Array Size : 484383744 (461.89 GiB 496.00 GB)

Confirm filesystem expansion:

# df -h /dev/md2
Filesystem      Size  Used Avail Use% Mounted on
/dev/md2        455G  134G  298G  31% /

For systems with multiple partitions in RAID1, you'll need to repeat this process for each array. The key differences between metadata versions:

0.90 metadata: Stores size in superblock
1.x metadata: Uses device size dynamically

If you encounter "device too small" errors during assembly, ensure:

1. Partitions are properly aligned
2. Both drives show identical sizes in /proc/partitions
3. No residual metadata from previous configurations

For automated solutions in deployment scripts:

#!/bin/bash
MDDEV=/dev/md2
PARTS="/dev/sda3 /dev/sdb3"

mdadm --stop $MDDEV
mdadm --assemble $MDDEV $PARTS --update=devicesize
mdadm --grow $MDDEV --size=max
resize2fs $MDDEV

Always backup critical data before array modifications. The --update=devicesize parameter is particularly important when:

• Migrating from smaller to larger drives
• Changing RAID levels that affect storage capacity
• Recovering arrays after partial failures

For reference, this solution applies to:

- Linux kernels 2.6.32+
- mdadm versions 3.2.5+
- Both ext3/ext4 filesystems