Case file · Logical & corruption · BDR-2025-0311
Cyclic Redundancy Check on My Life's Work.
Mid-use, a Toshiba external just simply became corrupt
: gone from File Explorer, and Disk Management would answer every attempt with a cyclic redundancy check error
. The owner did not need a translator to summarise his situation: life's work is on that hard drive
. And he genuinely did not know if the problem was in the hardware or the software.
Same symptoms on your desk?
01273 964902
The decode.
CRC errors show that your data came in. But, they also tell us that something went wrong when the accuracy of that data was verified. In other words, the drive has handed the data over, but it can't vouch for what it handed over. Two things could have caused this issue. Either you've got a messed-up file system spitting out junk from places on the drive where the actual data still exists, or you've got a disk surface beginning to fail. They're both bad but one is way worse than the other. The reason is that if you try to read a failing surface over and over again, the problem is going to grow. If you leave Windows running long enough by itself, it will keep attempting reads forever.
Equipment on this case.
How a job runs here →| The system | What it did on the day | Why this tool |
|---|---|---|
| DeepSpar Disk Imager 4 | Read the drive with retries capped, mapping weak sectors instead of fighting them | Head-map-aware imaging with per-head timeout, reset and power-cycle control |
| R-Studio Technician | Rebuilt the NTFS structures from the completed image | Broad filesystem coverage and dependable RAID re-assembly |
| ACE Lab PC-3000 Express | Confirmed SMART and defect-list state at firmware level before imaging began | Speaks the vendor-specific commands that sit outside the standard ATA set |
At the bench.
Establish which fault this actually is
Before any reads were performed on the drive, the initial checks were made using the firmware layer; specifically: SMART attributes, the growing list of defective blocks, and the history of re-allocation of said blocks. Based upon these results, it was obvious that we had a failing hardware device: the number of re-allocated sectors was increasing, and there were pending sectors present. It was not a confused file system, and the fixes for the two situations are vastly different.
Image once, gently, with the retries capped
This drive went onto the imager with tight timeouts and a hard retry limit. Regions that did not pass were logged for reference and skipped over instead of being hammered further, and then they were returned to at the end using a new read method. There is a very good reason for setting a maximum number of retries. On a degenerating disk surface, the fiftieth attempt at reading a bad sector is more likely to destroy the disk than to successfully read it.
Repair the filesystem where mistakes are free
Once an image of the entire drive had been obtained, all repairs occurred on the copy. The NTFS Master File Table had become corrupted in the damaged section of the drive; the records that survived were rebuilt and the few remaining "unreadable" sections were mapped so that we could tell the owner exactly what had been affected, rather than having to guess.
Where it ended up.
The life's work came off the image essentially intact, except for a handful of sectors. No critical files were lost. He also received a written document detailing exactly which sections had been lost. Before he paid a single penny, the recovery was verified against his priority folders.
Related on the index.
More from Logical & corruption.
Recognise your own drive in this story?
Same rule applies here as did with all previous scenarios: shut off the system first. Let the diagnosis be free before worrying about making any decisions regarding how to move forward.