A 30TB Mozaic Drive Isn't an Old IronWolf Case

Written and maintained by the Desert Data Recovery lab team. · Updated September 10, 2026

We already wrote about why helium still stalls recoveries, about Unraid Format and TrueNAS destroy-and-recreate, and about Synology Volume Crashed / Repair. This is not those posts again. This is the drive that is starting to show up in the same boxes: a Seagate Mozaic / HAMR member at 30 TB and up, wearing an Exos or IronWolf Pro label, failing in the middle of an array that the owner still thinks of as "just IronWolfs."

What you are looking at

Seagate's Mozaic 3+ platform is heat-assisted magnetic recording — HAMR. A laser on the write head heats a tiny spot so the drive can write at densities conventional magnetic recording cannot. The 30 TB-class disks shipping in the Exos and IronWolf Pro lines are that generation. So are the larger Mozaic capacities Seagate is putting behind the same badge.

From the NAS UI it looks like any other member: a bay went red, SMART threw a reallocated-sector count, the volume went unmountable, or the pool will not import. The sticker may still say IronWolf Pro or Exos. Capacity is the giveaway. A 4 TB, 8 TB, or 12 TB IronWolf you replaced last year was almost certainly CMR — perpendicular / conventional magnetic recording, a stack the industry has had tooling for years. A 30 TB Mozaic disk is not a bigger version of that part.

You will not see "HAMR" on the Synology, QNAP, Unraid, or TrueNAS screen. You will see the same Rebuild / Repair / Format language those UIs have always shown. That is the trap.

Why this is not last year's IronWolf case

Recovery on a mechanical Seagate is usually some mix of: stabilize the drive, match a donor head stack, work the service area / firmware, then image. On older CMR IronWolf, IronWolf Pro, and Exos families, that work is familiar. The heads are PMR heads. The media is PMR media. The servo and translator are families the major tools already know, or can be adapted from a close sibling.

HAMR changes the parts that job depends on:

• Heads. A HAMR head is not a PMR head with a new sticker. It carries a laser and a near-field transducer. You cannot pull an actuator from an older IronWolf Pro and expect it to fly and write on Mozaic media. A mismatch does not "almost work." It scrapes.
• Media. The platters are built for heat-assisted writes. Track layout, coercivity, and how a weak head damages the surface are not the same as on a 2018–2023 CMR disk. Imaging assumptions from those families do not carry over cleanly.
• Servo and firmware. The service area, adaptive data, and translator sit on a new stack. Tool modules written for last-generation Exos / IronWolf Pro do not automatically open this one. When the firmware side is missing, the case is research — the same stall we already describe on helium, plus a recording method those older modules were never written for.
• Donors. Matching a HAMR head stack means the same platform, not "another 30 TB Seagate from eBay." Early-generation parts are scarce. A lab that keeps a shelf of CMR IronWolf donors does not automatically have a Mozaic match.

Helium is still in the picture. These capacities are sealed, high-platter designs. Opening one is still a one-way extraction, and the aerodynamics still change the moment the seal breaks. HAMR does not replace the helium problem. It sits on top of it.

What a NAS owner should not do

Do not treat the rebuild window like last year's 8 TB swap. Rebuilding a 30 TB member is a long write. Hours become days. Every remaining disk is read hard for the entire pass. If a second member is marginal — common when the set was bought together — the rebuild is how you lose the redundancy you still had. That is the same write we warn about on Rebuild vs. recovery, only longer and on more expensive platters.

Do not keep writing. "Degraded but the shares are still up" is not a safe holding pattern on a 30 TB-class array. New writes go to a set that is already one failure from offline. If the failed member is still in the bay and still accepting commands, those writes can also land on a dying HAMR stack. Stop the volume. Then stop the disks.

Do not click Repair, Format, New Config, or destroy-and-recreate because the UI copied the language from a smaller array. Those buttons are covered in the Synology and Unraid / TrueNAS posts. They are writes. Capacity does not make them safer.

Do not shop a donor and open it. A used IronWolf Pro from the same bay size is probably the wrong generation. HAMR heads and older CMR heads are not interchangeable. Opening a sealed 30 TB disk on a bench that is not set up for helium is the helium failure mode we already documented. Doing it with the wrong actuator is both.

Do not run a generic HDD "repair" or firmware tool. Forum utilities written against older Seagate families can wipe a service area they do not understand. That is not a diagnostic. It is a write to the only copy of the adaptive data that drive has.

What recovery actually is

We do not "just swap the heads like an IronWolf." We do not start a NAS rebuild in the lab to see if parity comes back.

The job, when it is possible, is the same order we use on any failed array member: identify the platform, image each disk — including the one that dropped — and reconstruct from copies. If the Mozaic member needs mechanical or firmware work before it will image, that work is platform-specific. Donor matching, head work, and service-area access have to be valid for HAMR / Mozaic, not for the CMR family the badge used to mean. If the firmware or donor side is not there yet, we will say so. That is an evaluation answer, not a slogan.

We will not publish flash recipes, donor part numbers, or a list of which Mozaic model our tools already open. That is how the next drive gets scratched or wiped. We don't advertise a success rate.

There is no Mozaic or HAMR case in our public case log. We will not invent one. The Seagate jobs already on that page are older CMR and SMR families — Barracuda 3 TB members in a Synology RAID 5, an ST2000DM001, an ST1000LM035 laptop SMR disk. Different media. Different heads. Different tooling. This post is about the new platform, not a serial on the bench.

What to do right now

Do not rebuild. Do not Repair. Do not Format. Do not keep the shares online so the office can "finish the week."

1. Stop writes. Take the volume offline. If you cannot do that cleanly, power the NAS down. One more client save is not worth a second member.
2. Leave the disks in order. Do not reshuffle bays. Do not pull the red drive and put a new 30 TB in "to start the rebuild overnight." Label bays if you have to remove them.
3. Note what you have: NAS make (Synology, QNAP, Unraid, TrueNAS, other), RAID / SHR / ZFS layout, which bay failed, and a photo of the failed drive's label — model, capacity, any Mozaic / IronWolf Pro / Exos text. Photos are not a diagnosis.
4. Get the set evaluated. The technician on your case will tell you whether this looks like a Mozaic / HAMR member plus a NAS-layer problem, or something else. Free evaluation, no data, no fee.

If you are mailing it in, send every member, labeled, including the one you think is dead and any spare you already inserted. Pack them like any other failed drive — how to ship a failed drive safely. For the array-level work, see RAID data recovery. For the disk itself, hard drive data recovery.

If the array is Unraid or TrueNAS and the screen is about an unmountable disk or a pool that will not import, start with that post for the buttons — then treat the 30 TB member as this platform, not as a generic IronWolf. If DSM says Volume Crashed, that is the Synology post. If the only problem is a sealed high-capacity disk and you already know it is not HAMR, the helium write-up is the closer fit.