RAID & NAS · logged job · BHD-2025-9103
Four Members, and a Tender Due Monday.
The tender for a school extension closed at ten on the Monday morning, and the architects in Coventry had already worked most of it out for themselves. Their QNAP dropped off the network on the Thursday; by Friday it would not power on at all — it is the box that has died, not the drives
. All four were out and numbered in bay order before we touched anything
. What they wanted was a fixed price and a straight answer to one question: can you do it by Monday
.
Sounds like yours? Give us a ring.
0800 6890668
What that means.
Only one honest method fits a job like this into a weekend: run the stages over the top of each other instead of dropping any of them. Every member goes on an imager at the same time rather than waiting its turn, and the array layout is worked out from the copies that land first, while the rest are still running, so nothing sits idle waiting on something else. The shortcut we turned down was building the array straight off the practice's own disks. It saves a few hours, and it stakes the tender on four flawless reads; had one member faltered halfway through there would have been no copy underneath to fall back on.
What did the work here.
The order of the work →| Equipment | The job it did | What it adds |
|---|---|---|
| Atola TaskForce 2 | Copied all four of the members at once, and that is what saved the weekend | Every disk in the array goes through the imager together, not one after another |
| PC-3000 Portable PRO | Questioned each member for health and firmware state in place | Factory-level access out of a case that stands right next to the imagers |
| UFS Explorer RAID Recovery | Confirmed member order, stripe size and parity rotation from the finished images | Reads the volume layers a NAS puts over its array, not only the RAID underneath |
What we did.
All four on imagers together, none of them waiting
The courier came at midday and inside the hour every one of the four was on an imager and reading, and there they stayed until the last sector. Feeding disks in singly is what stretches a three-day array into a week. Take the whole set at once and Friday-to-Monday stops being Friday-to-an-apology.
Work out the layout while the copies are still running
The first images to finish carried metadata, and that supplied the candidate layouts: which member sat where, how wide the stripe ran, and which way parity rotated. Each candidate was then tested against real directory structures found on the volume, not waved through because that is how most controllers behave. By the time the fourth copy completed, the assembly was already down on paper.
Build it in software, check the tender before anything else
Built in software over all four of the images, the volume produced a filesystem that mounted at the first attempt. Checking then ran in the practice's order of priority rather than the disk's: the tender pack — drawings, specification, cost plan, appendices — went through first and was confirmed sound over the phone while the rest of the folders were still being validated.
How it closed.
The checked files reached the practice before Sunday ended, and on the Monday they filed the tender from their own drawings. Speed came from running the stages together and saying plainly where each one stood — not from leaving any of them out.
Other pages worth a look here.
The RAID & NAS cases.
Is yours behaving the same way?
Switch the device off, post it in, and decide nothing until the diagnosis comes back to tell you where you stand.