NAS & RAID · job record · NTG-2025-1491
Nine Months Between the First Failure and the Second.
Above the bench in a joinery works outside Worksop — the kind that does estate and listed-building work across the Dukeries — sat a four-bay Synology, and its fan had been drawing sawdust off the mitre saw for years. Since the turn of the year, bay 2 had been flagging bad sectors. Nobody rang anybody; the warning emails were simply switched off. Then, one morning, an order file refused to open. The unit was shut down and started again, and came back reporting volume failed
. Somebody had already written the plan on the lid in marker pen: fit a new disk and leave it to rebuild
. Of everything we were asked to do, that was the one thing we would not.
Yours doing the same? Say what it is up to — looking is free.
0115 8220606
What went wrong, and why.
The reserve in a RAID 5 set is exactly one disk. Parity lies thin, a little of it in each stripe, and it will rebuild a single absent member and go no further. Lose two and the arithmetic is left with more unknowns than it can solve. Powering the unit up for a quick look would therefore have cost this workshop twice over: two worn-out disks would have gone back to work, and a failed volume would have been free to lay fresh metadata over the structures a recovery depends on finding. It stayed unplugged. The marker-pen plan was refused in writing with the reasoning attached — put a failing disk into a live set and all you do is push more load onto the members still holding the thing up. No power went near it. The four disks were copied and every step from there ran against copies. From £500 + VAT is what NAS and RAID work is published at, and the assessment that sizes a job is free.
The bench work this one took.
How the work runs →| Kit on the bench | What it did on this job | Why we reach for it |
|---|---|---|
| DeepSpar Disk Imager 4 | Both failed members took new head stacks, the later failure imaged before the older | One platter side per pass, with the heads mapped before a single sector is taken off, and the resets, the timeouts and the power all held under our control |
| Atola TaskForce 2 | The two sound disks went onto imagers together while the failures waited their turn | Every member of a set copies in parallel, so no disk waits its turn behind another |
| UFS Explorer RAID Recovery | Read the mdadm superblocks first, then rebuilt the volume across four images | Works the stripe out again, then unpicks the volume manager the NAS laid on top |
What happened on the bench.
The pair that survived were tired too
Months of running a degraded array leave their own marks, and by this point both surviving members were showing pending sectors. Taking that pair concurrently on the TaskForce, ahead of the failures, saved several days — and more usefully, it meant nothing in this set was called healthy until it had been read right the way through.
What was in the event log
Both failed disks read once a donor head stack was in each. Then the event log altered the whole shape of the job. That noisy bay had dropped out of the array back in January, leaving its contents nine months stale — a portrait of the volume taken well before any of this began. The recent material was all on the disk that failed last, and there the unreadable regions formed bands that could be worked around.
Three images used, January discarded
Member order, stripe width and the direction the parity rotates in were all taken from what the array had written about itself; nothing was assumed. The assembly then used three recent images and dropped the turn-of-the-year disk entirely, on the grounds that blocks written months apart can produce corruption that reads as consistent and is not. What the newest disk had held back turned out, in the event, to lie in unallocated space.
What was returned.
First attempt, and the volume mounted. Every drawing, job card and order book was set against the workshop's own numbering before replacement disks and new media were posted back. The Synology has moved off the shelf above the bench and into the office, and its alert emails are on again. The figure had been settled in writing before a disk was opened.
Start here if the same thing is happening to yours.
Other RAID & NAS jobs, written up.
Sounds like yours? Switch it off before anything else.
£250 + VAT covers a card or a USB stick; a single drive is £300 + VAT; RAID and NAS start from £500 + VAT. Nothing is charged for the first look, and a fixed figure reaches you in writing before any chargeable work begins. Switch the device off, leave it alone, and post it in. On most jobs, no data back means no fee.