Data Recovery Case File · NAS & Network Storage · Two Chassis, One Volume

Ten Years of Uptime, Ended by Maintenance

His enquiry describes conscientious ownership producing exactly the wrong result. A five-disk network unit with a second five-disk expansion cabinet attached, run continuously for a decade: "apart from individual disk failures from time to time, the system has operated 24/7 for 10 years. Every so often I shut down the system and remove the disks so that I can clear both units of accumulated dust. After doing this two days ago, the system will now not switch on." He does everything right and this still happens. Two things need separating here — why a planned shutdown ends a decade of service, and why an array spread across two chassis needs more care than one.

The decode: why maintenance did it, and the two-chassis problem

Why equipment that runs continuously dies at power-on: a device running for years reaches a stable thermal and electrical state and simply stays there. The moments of real stress are starting — capacitors that have degraded quietly under a decade of heat are asked to handle inrush current, the power supply delivers its largest surge, and ten drives spin up from stationary at once, which is the most current-hungry thing any of them ever does. A unit that has been up for years has been ageing invisibly the whole time, because nothing ever asked it to start again.

Why his careful maintenance is not to blame: and this is worth saying plainly, because it feels like it is. Clearing dust from a unit run continuously for ten years is good practice — dust is a real cause of thermal failure, and he has almost certainly extended the machine's life by doing it. The restarts were always going to be the risky moment, and doing them deliberately, on a chosen day, is far better than discovering the same fault during a power cut. The maintenance revealed the condition; it did not cause it.

What has most likely failed: in order of probability, and none of it involves the data. The external power supply, which is the commonest fault on these units and costs nothing to establish. The unit's own power circuitry, where degraded capacitors are the well-known age-related failure at ten years. Or one or more drives failing to spin after a decade of continuous rotation, which is the same phenomenon at disk level.

Now the part specific to his configuration, and it is the one that could turn a repairable situation into a difficult one. His volume spans two separate chassis. That means the member order is not simply bays one to five — it is bays one to five in the main unit and bays one to five in the expansion cabinet, as a single ordered set. Having removed every disk from both units for cleaning, the order matters enormously and is easy to disturb. A disk returned to the wrong bay, or worse to the wrong chassis, changes its position in the volume. Nothing should be powered until the original arrangement is established, and if there is any doubt about which disk came from where, that uncertainty should be stated rather than guessed at — because member order can be determined from the disks themselves, and a wrong guess acted on by the appliance is much harder to undo.

What must not happen: no repeated power-cycling, and no rebuild if the unit does come up degraded. The link between the two chassis is also worth checking as a separate failure point.

On the bench

Supply and appliance electronics were eliminated separately before any disk was implicated, since a unit that fails for want of a power supply should not be quoted as a recovery. Members were removed and labelled by chassis and bay — not merely by bay, since a volume spanning two enclosures has an order that runs across both and a disk returned to the wrong chassis changes its position in the set. All ten were imaged individually write-blocked on the Atola TaskForce 2, and the volume assembled offline from the images, where member order could be determined from the disks themselves and no appliance could attempt a rebuild.

The outcome

The supply and appliance electronics eliminated first, every member imaged individually with chassis and bay recorded, and the volume assembled offline. Free assessment, one fixed written figure including VAT; where a drive has to be opened, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The decode, for anyone whose always-on unit died after maintenance: the cleaning revealed the condition rather than causing it, because equipment running continuously is stressed most at power-on and a unit up for a decade has been ageing invisibly. Check the power supply first. And where a volume spans two chassis, label by chassis and bay — order runs across both units, and a disk returned to the wrong one changes its position.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.