Can You Expand a Mirrored Array With Larger Drives One at a Time?

You can often expand a two-disk mirror by replacing one drive, completing the rebuild, then replacing the other. Capacity usually remains unchanged until every mirror member is larger and the array is explicitly expanded.

The method is common but not universal. The controller or filesystem must support replacement with larger members and later growth of the virtual disk, partition, pool, and filesystem. A verified backup and a clean starting mirror are prerequisites, not optional precautions.

The Conditional Answer Depends on the RAID Layer

Hardware RAID, mdadm, ZFS mirrors, Btrfs RAID1, and appliance-managed storage expose different controls. Some automatically recognize extra member capacity; others require a separate virtual-disk expansion, partition resize, pool operation, or filesystem grow step.

Confirm the exact implementation and version before touching hardware. The safe high-level pattern is replace one disk at a time and allow a successful rebuild before changing the next source of redundancy.

Start With a Clean Mirror and a Restorable Backup

Do not begin an expansion while the mirror is degraded, scrubbing with errors, or showing unstable links. Read every member’s SMART data, complete a consistency check, map serial numbers to bays, and confirm the backup can restore the most important data.

Expansion causes at least two full member reads and writes. That workload can expose weak sectors that normal use has not touched. If the surviving old drive reports pending or uncorrectable sectors, copying data to a new array can be safer than rebuilding from it twice.

Replace the First Drive and Wait for Full Redundancy

Install a drive whose usable sector count is at least as large as the old member. Matching the marketing capacity is not enough because models with the same label can expose slightly different numbers of sectors.

A slightly smaller replacement may be rejected even when both products are sold as the same nominal size. Check exact logical sector count, sector format, interface support, and controller compatibility before ordering.

Verify the First Rebuild Before Replacing the Second

Wait until the new member is active and the mirror is healthy. Save the final status, run a short health review, and check that read, write, link, and media errors did not increase. Do not treat 100 percent on a progress bar as the only completion signal.

The larger-drive mirror migration pattern requires each replacement to rebuild fully. Pulling both original members removes the live mirror and turns an in-place expansion into a restore or recovery exercise.

Replace the Second Drive Using the Same Controls

Once the first new member is fully synchronized, replace the remaining smaller disk and repeat the same checks. Keep the first old disk untouched until the second rebuild and a post-rebuild integrity check finish, because it can still be valuable during recovery.

Avoid changing partitions or growing filesystems between the two replacements unless the platform explicitly requires it. The array will normally continue to expose the old size while any member still defines the smaller boundary.

Extra Capacity Is a Separate Expansion Step

After both larger drives are healthy, inspect the storage stack from bottom to top. The controller may show unused physical space even though the virtual disk remains the original size. Then the partition table, logical volume, and filesystem can each have their own unchanged limit.

A practical unused capacity after replacement example shows why replacing disks and exposing space to the operating system are separate operations.

When Not to Use the One-at-a-Time Method

Do not continue when the controller cannot expand the virtual disk, the mirror uses unsupported sector formats, the remaining source drive is deteriorating, or the appliance documentation requires migration to a new pool. A fresh array and verified restore may produce less risk and downtime.

Also stop if the first rebuild creates new I/O errors. The procedure is intended for a healthy mirror that is being enlarged, not for extracting data from unstable members.

Plan for Two Full Rebuild Windows

Sequential replacement is not one maintenance event but two separate periods of reduced redundancy. Reserve enough time for each rebuild, its health review, and a pause between them so an error on the first new member is not carried into the second change.

Keep normal workloads light during both passes. If the expansion is capacity-driven rather than urgent, waiting for a low-demand window is safer than combining drive replacement with unrelated updates, cable changes, or filesystem maintenance.

FAQ

Will the mirror use the larger drive’s full size immediately?

Usually not. While one smaller member remains, the mirror generally uses only the common capacity. Even after both replacements, a separate array and filesystem expansion may be required.

Can I use different brands or speeds?

Often yes when interface, sector format, capacity, and controller rules are compatible. Performance and error-recovery behavior may follow the slower member, so matching operational characteristics is preferable.

Should I keep the old drives?

Keep each original drive unchanged until both rebuilds, the expansion, a full integrity check, and a backup verification finish. Label them by serial number and removal order.

The Safe Boundary

Replace only one mirror member at a time, require a healthy rebuild before the next change, and grow each storage layer only after both larger members are active and verified.