mdraid reshape recovery

An interrupted mdadm reshape can leave one array containing both old and new RAID geometry.

mdadm reshape changes member count, RAID level, chunk size, or layout while moving data. If it stops, the reshape position separates regions written with the new geometry from regions that may still use the old one.

Understand the symptom

The same logical array can require two decoding rules during an incomplete reshape.

WHY

What is happening

Kernel md metadata can retain new and old values for raid_disks, chunk_size, and layout while reshape_position records progress. Backup-file data may protect a critical stripe window. Recovery maps member roles, events, data offsets, geometry, and boundary state before reconstructing either side of the reshape.

What you may see

Symptoms that change the recovery path.

01

Array will not resume reshape

Assembly cannot reconcile member events, geometry, or the saved progress position.

02

Wrong device count reported

Metadata contains both old and new member counts during transition.

03

Filesystem corrupt after forced assembly

One geometry was incorrectly applied across both sides of the reshape boundary.

04

Backup file requested

mdadm needs the external critical-section copy created for the reshape.

Protect the source

Actions that can make the situation worse.

01

Force assembly can advance wrong metadata

A writable start may commit an incorrect member set or geometry.

02

Zeroing superblocks removes transition evidence

Old and new layout, role, event, and reshape records can be lost.

03

Restarting reshape overwrites boundary stripes

The critical window may include the only intact version of moved data.

Do this now

Safe steps before evaluation.

01

Stop array starts and writes

Disable boot assembly, reshape resume, fsck, LVM activation, and application access.

02

Label and image every member

Preserve failed, removed, spare, replacement, and active devices in original slot order.

03

Save metadata and history

Capture mdadm --examine, sysfs geometry, reshape position, data offsets, logs, and exact original command.

04

Locate the backup file

Preserve any --backup-file and the filesystem or initramfs where it was stored.

NEPFOSYS process

How a data-first evaluation proceeds.

01

Dual-geometry mapping

NEPFOSYS reconstructs old and new layout, roles, offsets, events, and reshape boundary.

02

Member and backup correlation

Images and critical-section backup data are tested without writing to originals.

03

Virtual array validation

File systems and priority data are checked across both geometry regions before extraction.

A clear next step

Preserve the original device before another repair or recovery attempt.

Start a case with the exact symptoms, incident timeline, prior attempts, encryption details, and the files that matter most. NEPFOSYS will route the case to the appropriate DFW intake or nationwide mail-in path.

Start with a case review

Questions, answered

What customers ask before starting.

What is mdadm reshape_position?

It records how far the array has progressed through a geometry-changing reshape.

Why can old and new values both appear?

During reshape, md tracks the geometry already written and the geometry still needed for the remaining region.

Can I restart the reshape without its backup file?

That can be unsafe when the protected critical section is needed. Preserve all members and metadata before any restart.

DFW local · Nationwide mail-in

Protect the device. Start with the facts.

Start my recovery case