Data Recovery Case File · NAS & Network Storage · Bought the Disk, Never Fitted It

The RAID Error He Meant to Act On

His enquiry contained the most human sentence in this volume, tucked into the middle of a technical description. A two-bay network drive, out of support, showing what he had seen described as the "yellow light of doom" — powering up blue, ethernet flashing, then turning amber and stopping. And then: "the drive has not been used for some time, as I had noticed a RAID error in the logs and had purchased a replacement drive with the intention of moving the data, but since..." He saw the warning. He understood it. He bought the replacement. And the migration never happened. This page is about the technical position — and about the gap between recognising a window and using it, which is where most of this archive actually lives.

The decode: the light, the appliance, and the window

What the status light is describing: these units use a small set of colours to report their own state, and the sequence he describes — normal on power-up, then amber, with networking continuing — is the unit starting successfully and then failing at the point where it tries to bring up its storage volume. The networking staying active is the tell: the appliance itself is running, it is the array underneath that it cannot assemble. That is consistent with the RAID error he had already seen in the logs, now progressed.

Why being out of support does not matter: as with any orphaned appliance, the disks inside are ordinary drives carrying a Linux filesystem. The manufacturer withdrawing support removes the intended route in and touches nothing on the platters. The disks come out and are read directly. There is one trap that goes with it: connect one to a Windows machine and it will show as unformatted and offer to initialise it, which on an array member destroys the layout or a complete copy depending on the configuration.

What must not happen now: the unit must not be left running or restarted repeatedly, and above all it must not be allowed to rebuild. A rebuild reads every remaining member intensively — it is the most punishing operation an array performs and precisely when a second marginal disk fails. Nor should the replacement disk he bought be fitted now: inserting a new member into a degraded array is what triggers exactly that rebuild. The disk he purchased has a use, and it is as a destination for recovered data rather than as a repair.

The window, said kindly: he did the hard part. He read his logs, understood what an array error meant, and spent money on the fix. What did not happen was the migration itself — the several hours of copying that stands between intention and safety. That gap is not carelessness; it is the ordinary shape of a busy life, and it is why so many of these enquiries begin with I knew about it. The practical lesson is narrow and worth stating: when an array reports an error, the first action is copying the data off, not buying a disk. Replacement hardware protects the future; a copy protects the past. If only one gets done, it should be the copy.

On the bench

The unit was powered down before anything else, because an appliance repeatedly attempting to assemble a degraded array is doing the one thing most likely to cost a second member. Both disks were removed, labelled by bay, and imaged write-blocked on the Atola TaskForce 2 — both rather than one, since the array's parameters are derived from the members and a degraded array's surviving data may need either. The array was assembled offline from the images, where no appliance could attempt a rebuild and no host could offer to initialise anything, and the Linux volume was read from the copies. The contents were verified and delivered on cross-platform media.

The outcome

The unit stopped before it could rebuild, both members imaged and the array assembled offline from the copies. 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 array has warned them: the status light changing after a successful start-up, with networking still active, means the unit is running and cannot assemble its volume — which is a logged array error progressed; being out of support is irrelevant, because the disks inside are ordinary drives with a Linux filesystem; do not let it rebuild and do not fit the replacement disk now, since inserting a member into a degraded array triggers exactly the operation that kills second disks; and when an array reports an error, copy the data off first and buy hardware afterwards.

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.