Taking work now — the first look is freeWhole sets posted in from anywhere in the UK, or handed in at ten drop-off pointsQuicker still, give us a ring:0800 6890668
RARRAID Array Data Recovery 0800 6890668 Price my job
RAR / Controllers and hosts / Linux mdadm

mdadm · md0 · /proc/mdstat · --examine · --detail · superblock 1.2 · left-symmetric · Ubuntu · Debian · Proxmox

Linux mdadm array recovery. The software RAID under most of the world's NAS units, and the most honest log of what happened.

mdadm manages the Linux md driver, and md is under more arrays than any controller: every Synology and QNAP, most home and small-business Linux servers, Proxmox hosts that chose it over ZFS, and appliances that never mention it. It keeps a superblock on every member with the array's identity, geometry and each member's role and event count, in one of four places depending on the metadata version, and it assembles from the members whose counts agree, logging exactly what it did in dmesg. The states are in the mdadm and levels pages; this page is about the tool and the platform: which commands are read-only, which write, what --examine prints and why every line of it matters, and how a set that Ubuntu, Debian or RHEL will not assemble is assembled on the bench from images. A set of two to four members is £500 + VAT upwards after the free look, fixed in writing, 5–10 days at the bench.

Free first lookOne fixed figure in writingNo data, no bill on most jobsReturn postage paid

Rather talk it through? An engineer answers the bench line
0800 6890668

Power down. Label every slot before a drive comes out. Do not rebuild, do not force a drive online, do not import or clear a foreign configuration, do not initialise, and do not run chkdsk, fsck, zpool import -F or mdadm --create on the members. A rebuild reads every sector of every survivor and writes to the replacement; each of the commands writes to the drives. On images, all of them are reversible. On the originals, none of them is.

Models and versions we see.

Which level is it →
Family Models or versions Metadata, defaults and notes
Read-only commands--examine, --detail, --examine --scan, --assemble --readonly, cat /proc/mdstat, dmesgCopy their output; they write nothing
Writing commands--create, --assemble --force, --add, --re-add, --grow, --zero-superblock, --runNever on originals
Metadata 0.90Old arrays; superblock near the endSurvives accidental partitioning of the start
Metadata 1.2Modern default; 4K from the start; data offset 128MiBA --create changes the offset
ContainersIMSM (Intel), DDF (Broadcom)md127 container, md126 volume

What tends to go wrong on mdadm.

Which commands write--examine, --detail and --assemble --readonly are read-only. --create writes superblocks, --assemble --force promotes a stale member, --add and --re-add start a resync, --grow starts a reshape, --zero-superblock erases the description. The forum offers the writing ones.
The data offsetMetadata 1.2 leaves a gap between the superblock and the data, and its size has changed with mdadm versions, to 128MiB on modern ones. An array created years ago and re-created today with the same parameters has its data start in a different place; the superblock says one thing and the file system sits elsewhere.
Interrupted reshapes--grow from four members to five, or from RAID 5 to 6, rewrites the array in place with a backup file for the critical section. Interrupted by a power cut, it leaves part of the array in the old geometry and part in the new, with the reshape position in the superblock. The bench assembles each part with its own geometry.
The NAS underneathSynology, QNAP, ReadyNAS and Buffalo all run md, with LVM on most of them. Their dashboards are the NAS site's subject; the md arrays are read here from the same superblocks with the same tools.

Linux mdadm symptoms, and how long each gives you.

Not listed? Describe it on the form →
Degraded; a member missingStill readable. Image the missing member before anything rebuilds.Do not delay
Will not assemble, import or attachThe software is being careful. Nothing written yet.Do not delay
A forum command was triedSay which, in order. It decides the plan.Do not delay
Corrupt after a power cutTorn writes; the newest consistent state on images.Everyday job
Files wrong after an updateA metadata format change. Do not repair on the originals.Everyday job
Packing it and posting it: power down, photograph the controller's screen or export its log, and write the slot number on each drive with a marker before it comes out of its carrier. Send every member of the set, including the one that failed first; it holds the stripes the rebuild never reached. Each drive travels in an anti-static bag inside its own padding, in a box with nothing able to move. Send the controller if it was replaced or holds an encryption key; otherwise it stays. Insure the parcel for what the data is worth, use a tracked service, and know that the posting address is not printed anywhere on this site; it arrives by email in reply to the form, with a booking sheet to print; the sheet inside the parcel is what matches it to your enquiry when it is opened. Or hand the sealed parcel in at the nearest of ten drop-off points, your name on the outside and the sheet inside; say where you are on the form and it comes by email. We pay the postage home either way. The whole of it is written up on the guide to packing and posting.

What the message means on mdadm.

Describe yours to us →
What you see The usual reason Where that leaves you
Reshape interrupted; array will not assembleTwo geometries on the membersAssembled in two parts from the reshape position
--grow ran out of space or backup file lostCritical section not backed upAssessed for the stripes in the critical section
DegradedA member out; still readableImage the weak member before any resync or resilver
Will not assemble or importToo few current members, or metadata damagedMetadata read from images; assembled in software
Read-only or suspendedErrors stopped I/ORead-only on images at the newest consistent state
Command already run--create, -F, chkdsk, fsckAssessed for what was written

From the parcel arriving to your files going back.

Work we have closed →
01

Logged the day it lands, and the first look costs nothing Free

A case number goes on the parcel and a number on every member the day it is opened, matched to the slot you wrote on it. Each member goes on the imager its interface needs, never on a controller, and its metadata is read before a sector is: the level, the order, the stripe size, the event counts that say which member is current and which dropped out first. An engineer settles what has happened to the set and how much of it can honestly be read back. Back to you come two things together: a straight note of what is liftable and what is not, plus one figure, fixed and written down. Accept it, or decline and owe us nothing.

Nothing to pay for lookingA single figure, put in writingNo rebuilds, no imports, no initialise
02

Every member imaged, including the one that failed first

Every drive in the set is imaged sector by sector, weak areas last, on hardware that controls every retry, with a map of what could not be read kept for each. Members with failed heads go to the clean bench first; that is the drive site's work and the same bench does it. The drive that dropped out first is imaged too, because it still holds every stripe written before it dropped, and a rebuild that stalled part-way never reached them.

Sector by sector, weak areas lastThe first-dropped member included
03

The geometry, from the metadata or from the parity

Where the controller's metadata survives on the members, the order, stripe size, parity rotation and data offset are read from it. Where it was cleared or overwritten, they are recovered from the data: parity across the members at the same offset should XOR to zero on a consistent stripe, which confirms the level and finds the stale member; where parity lands says the rotation; entropy at stripe edges and the file system's own anchors give the stripe size, the order and the start.

Metadata first, parity secondOrder, stripe, rotation, offset
04

Assembled in software, and repaired on the virtual volume

The set is put together from the images in software, with nothing written to any of them: the current members in, the stale member used only to fill holes a survivor could not give. The file system is checked and repaired on a copy of the virtual volume, VMFS and CSV volumes opened and the virtual machines' disks extracted, databases repaired where they need it. The originals are not touched again.

Nothing written to the imagesVirtual machines and databases opened
05

You see the file list before you pay

What was recovered is listed for you first, and only then does a bill exist. Approve the list and it is invoiced; turn it down and it is not — and where nothing has come back, most jobs carry no charge at all. Recovered data travels home on fresh media bought in for your job, with the postage at our end. Your case is not closed until you have opened the files on a machine of your own.

No charge until you accept the figureFresh media, supplied with the job5–10 days at the bench

From the bench

  • Copy out --examine for every member, /proc/mdstat and dmesg first. They are the case, and they write nothing.
  • Say the mdadm version and the distribution. The data offset and the default chunk changed with them.
  • An interrupted reshape is recoverable from the reshape position in the superblock; do not restart it.
  • Copy out the read-only diagnostics first: --examine, zpool status, Get-VirtualDisk, dmesg. They are the case.
  • The forum's fix writes. --create, --assemble --force, zpool import -F, chkdsk and fsck all belong on images.
  • Say the version. mdadm, OpenZFS and Windows all changed their on-disk formats, and the tools follow.

One job, followed all the way through.

UK · RAR-2026-0630JOB LOGGED ✓

A Proxmox host's five-drive mdadm RAID 5 mid-reshape to six members when the power failed, the array refusing to assemble afterwards, and the backup file on a volume that was on the array

All six drives came in. The superblocks recorded the reshape position, and the array was assembled from the images in two parts: the stripes before the position in the six-member geometry and those after it in the five-member one, with the critical section reconstructed from parity across both. The ext4 volume mounted read-only and the virtual machines' disk images were copied out.

99.9% of the volume recovered9 days here, and back by post
Illustrative example — replace with a genuine case

What helps, and what harms.

Do this much first

  • Copy out the read-only diagnostics
  • Power down and label the members
  • Send every member, and any cache, log or witness device
  • Tell us the version and every command that was run

What sets us back

  • mdadm --create or --assemble --force
  • zpool import -F, -X or -T
  • chkdsk or fsck on a degraded or mis-assembled volume
  • Resetting or recreating the pool

Questions answered before you commit.

Which mdadm commands are safe to run on a failed array?

--examine, --detail, --examine --scan and --assemble --readonly, and reading /proc/mdstat and dmesg. Everything else writes: --create, --assemble --force, --add, --re-add, --grow, --zero-superblock.

Where is the superblock?

It depends on the version: 0.90 and 1.0 near the end of the device, 1.1 at the start, 1.2 four kilobytes from the start. --examine tells you; the bench looks in all four places.

My reshape was interrupted by a power cut. Is the array gone?

No. The superblock records the reshape position, and the array is assembled in two parts, each in its own geometry, with the critical section rebuilt from parity.

What does it cost?

£500 + VAT upwards for a set of two to four members after the free look, fixed in writing; larger sets from £1,250 + VAT.

Begin here if yours is doing the same thing.

mdadm: not enough devices, array inactive →mdadm and LVM →SHR and Synology RAID F1 →Intel RST, RSTe and VROC →

Nothing gets worse while it is powered down.

Looking at it is free. Back comes a list of what opened and what did not, together with a single price to finish, set down in writing while you are still free to say no. On most jobs an invoice only follows the data. Until that list reaches you, leave the server off and the drives in their slots.

0800 6890668

Price my jobPacking and posting