Message first
What the controller is saying, and what to do tonight. The dialects differ. The first move does not.
A RAID controller says what is wrong in its own dialect: a PERC says Virtual disk degraded and Foreign configuration found, a Smart Array says Logical drive failed and Predictive failure, MegaRAID says Unconfigured bad, mdadm says kicking non-fresh. The drive itself says it more plainly: it clicks, it vanishes from the bus, it reports 0 GB, it reads as noise. Underneath, they describe a handful of things, and the right first move is the same for every one of them. Find yours below. One member drive imaged is £300 + VAT, a set of two to four is £500 + VAT upwards, fixed in writing after the free look, and on most jobs no data means no bill.
Rather talk it through? An engineer answers the bench line
0800 6890668
What the controller says, or what the drive is doing.
Rather begin with the kind of drive →What the controller says
PERC virtual disk degraded or failedDegraded: one member out, still readable. Failed: too many. Both: image the weak drives before anything rebuilds→Foreign configuration foundMetadata the controller does not claim. Do not clear it; import only if every member is present→Punctured stripe on a Dell PERCA rebuild with errors that kept the array up and the lost stripes lost. What can still come back→HPE logical drive failed or predictive failureSmart Array's two verdicts, and which of them means power down tonight→SSDs failed at 32,768 or 40,000 hoursHPE and Dell SAS SSDs with a clock in the firmware. What survives, honestly→mdadm: kicking non-fresh, read errorLinux software RAID's log lines, what each means, and the two commands never to run→What the drive is doing
RAID drive clicking or spinning up and downFailed heads, or a service area that will not load. Power down; every start scratches→SAS drive not detectedBackplane, cable, firmware or the drive. The checks that are safe, and where they stop→Drive shows 0 GB or incompatible sector sizeA 520-byte or 4Kn drive on the wrong host, or a firmware failure. Do not initialise→SED drive lockedThe key stayed on the old controller. Find it; never PSID revert→Predictive failure, SMART and grown defectsWhat the thresholds mean, which attributes matter, and when to image rather than replace→Several drives failed at onceA backplane, a supply, a controller, or a firmware clock. Usually not four dead drives→The one rule under all of them.
Every message on this page is the controller's way of saying that a drive answered too slowly, stopped answering, or is describing a set the controller does not recognise. None of them means the data is gone. What loses the data is what happens next: a rebuild that reads every sector of survivors bought on the same day as the drive that failed, a foreign configuration cleared to make the warning go away, a drive initialised because the host could not read its sector size, a PSID revert on a locked drive, a firmware update pushed onto SSDs that had just died of firmware.
So the first move is the same whatever the dialect. Export the controller's log if it will still give one. Power down. Write the slot number on each drive before it comes out, and send the weak drives, or the set. The bench images every drive on the equipment its interface needs, at its native sector size, and either hands the image back on fresh media for the rebuild or reassembles the set from the images in software, where nothing can be lost by trying. The pages above say the same thing in each controller's own words, and add what its own manual leaves out.
The questions that come up first.
The set is degraded but still working. Should I rebuild?
Only if every surviving member is healthy and you have a backup. If a second drive shows predictive failure, the rebuild reads the bad sectors that finish it. Copy what matters, power down, and image the weak drives first.
Should I import the foreign configuration?
Only if every member of the set is present and nothing was written since it was last consistent. If in doubt, no. Clearing it is never the answer.
The controller says failed but the drive spins. Is it dead?
Usually not. Enterprise firmware gives up on a bad sector in seconds so the controller can carry on; the controller drops the drive; the drive reads on a bench that waits.
What does it cost?
One member drive is £300 + VAT; a set of two to four is £500 + VAT upwards; larger sets, 15K SAS sets and shelves from £1,250 + VAT. Fixed in writing after the free look.
Not on the list?
Copy out what the controller says, or export its log, and send it with the form. An engineer places it faster than any menu, and asking is free.