My hard drive started making strange noises and now it has bad sectors. I can still see some files, but copying them fails or freezes my computer. I have important photos and work documents on it, and I need advice on the safest way to recover data from a failing hard drive before it gets worse.
Stop poking the bad drive
If your drive starts spitting CRC errors, freezes when you open folders, or stalls mid-read, stop right there. I would not keep testing it. I did this once out of habit, kept retrying, and the drive got worse within the hour.
If it is external, unplug it. If it is your internal system drive, power the machine down. Every extra read attempt puts more stress on a disk that is already failing. The drive keeps trying to pull data from bad spots, and the OS might also start its own repair routines in the background. Bad combo. You lose ground fast.
Two kinds of failure, and they do not behave the same
Sometimes the damage is logical. I mean the file system or sector data no longer matches its error correction data. This often shows up after a crash, power loss, forced shutdown, stuff like that. The hardware itself might still be fine. People call these soft bad sectors.
Then there is physical damage. Hard bad sectors. This is where the media is damaged, or the heads are wearing out, or something inside the drive is going off the rails. No software fixes physical wear. It only works around the damaged areas, if the drive stays alive long enough.
Why I would skip CHKDSK at the start
A lot of people go straight to CHKDSK. I would not, not first.
CHKDSK is meant to make a file system usable again. It is not a data rescue tool. If the file system is broken, it may clean things up by cutting off damaged references and orphaned entries. Nice for booting a machine. Bad for recovery odds. You end up with a neater disk and fewer paths back to your files.
The safer move is to avoid changing the source drive at all. Make a sector-by-sector copy onto a healthy disk. Work from the copy. Leave the original alone after imaging.
The home recovery route I would take
If you are doing this yourself, recovery software with imaging support is the thing to look for. Disk Drill gets mentioned a lot for a reason. One useful part is the byte-to-byte backup option.
What matters more is how the software behaves when it hits unreadable spots. Older tools, and some current ones too, tend to get stuck hammering the same bad area. That wastes time and puts extra strain on a weak drive. Better tools skip past trouble, copy the readable parts first, then return later and retry the bad sections in smaller chunks. I have seen this make the difference between getting most files and getting a half-dead clone.
The sequence I would follow
- Get a healthy destination drive with enough space.
- Create a full image or clone of the failing drive.
- Disconnect the original once imaging is done.
- Load the image into your recovery tool.
- Scan the image, not the damaged disk.
- Save recovered files to a different healthy drive.
Do not write recovered files back onto the failing disk. I know this sounds obvious, but people do it when they are low on space. It makes a mess.
When DIY stops making sense
If the drive is clicking, grinding, beeping, refusing to spin up, or not showing up at all, I would back off. Those signs usually point to mechanical failure. At that stage, home tools are a gamble.
If the data matters, a lab is the safer route. Places like Gillware or Techchef handle this kind of thing in controlled environments and use hardware most of us do not have. They swap parts, stabilize failing drives, and pull data with specialized gear. It is not cheap. Expect something like $500 to $3,000 depending on what failed. Painful, yeah. Still cheaper than losing family photos, tax records, or client files you cannot replace.
What to do with the drive after recovery
Once your files are recovered and you have checked them, opened them, and made sure they are not corrupt, then you can test or wipe the original drive if you want. A full format or chkdsk /r will mark bad sectors so the system avoids them later.
I still would not trust the drive again for anything important.
A drive that has started doing CRC errors and hanging reads has already told you what it is. Retire it from real use.
The part people skip until it hurts
The long fix is backups. Plain old 3-2-1.
- Keep 3 copies of important data
- Use 2 different storage types
- Keep 1 copy off-site or in the cloud
I ignored this for too long. Then one bad drive ate a weekend and shaved a few years off my mood. Backups are boring. Recovery is worse.
1 Like
Strange noises changes the math. This is not a cleanup job first. It is a failing hardware job.
I agree with @mikeappsreviewer on one big point, stop normal file copying. Where I differ a bit is this. If the drive still mounts and the noise is light, I would first grab the irreplaceable stuff only, in short bursts, before running any broad scan. Photos, docs, client files. Skip huge videos and system folders. A full scan pounds the disk for hours. Sometimes you do not have those hours.
A few practical options.
-
Pull SMART data once.
Use CrystalDiskInfo on Windows or smartctl on Linux/macOS. Look at Reallocated Sectors, Pending Sectors, Uncorrectable Sectors. If Pending or Uncorrectable is climbing, the drive is dying fast. If it freezes during SMART read, that is bad news too. -
Copy by file type, smallest first.
Start with Documents and Pictures. Small files often come off before the drive gets worse. Use TeraCopy or FastCopy with retries set low, like 1 to 2. Default Windows copy tends to hang too long. -
Skip USB if possible.
If this is an external HDD, remove it from the enclosure and connect by SATA. A lot of cheap USB bridges handle read errors badly and freeze the whole session. Direct SATA is often more stable. -
Cool it down.
Not freezer nonsense. I mean a desk fan, stable surface, no movement, no bumping the drive. Heat pushes weak drives over the edge. -
If you need software, use Disk Drill on a healthy machine and save output to another drive.
Disk Drill is solid for recovering deleted or damaged-file-system data after you get what you can. I would not point it at a drive making loud clicks for an all-night scan. If the disk is still readable, target folders first, then scan if needed. -
If the sound is clicking, chirping, or spin-up spin-down, stop DIY.
At that point, every retry hurts your odds.
One more thing. This page explains bad sector tables in plain english, useful if you want the low-level side of a bad block error without too much fluff:
how hard drives track and remap bad sectors
Short version, if the data matters more than the drive, spend your remaining reads on the most important files first. Then retire the disk. Do not trust it agian.
Strange noises + freezes is where I slightly disagree with both @mikeappsreviewer and @sonhadordobosque on one thing: I would not spend too much time deciding between file copy first vs full clone first if the drive is audibly getting worse. That decision window can close fast.
My take:
- If the data is worth real money, stop DIY and send it to a lab now.
- If pro recovery is not happening, do one short triage pass only.
- Grab the truly irreplaceable files first if they are easy to access.
- If access is flaky, switch immediatley to imaging.
What I would add that has not been said much yet is to avoid booting from that drive at all. If this is your main Windows drive, remove it and attach it as a secondary drive to another PC, or boot from a Linux USB. Your own OS thrashing pagefile, indexing, thumbnails, antivirus, and automagic repair junk can make a bad situation worse.
Also, do not trust SMART too much just because it says ‘Good.’ Look at the actual numbers for pending, reallocated, and uncorrectable sectors. This is a decent explainer on how to read SMART bad sector counts before a drive dies.
If you go DIY software route, Disk Drill is fine, but I would use it after stabilizing access or after making an image, not as the very first panic click-everything move. Recovery tools are helpful, but a dying mechanical drive does not care about your software optimism lol.
One more thing people forget: if the drive is making scraping, repeating clicks, or spin-down noises, do not open the enclosure and do not tap it, freeze it, or ‘let it rest overnight’ like some 2009 YouTube wizard said. That stuff kills drives.
Short version: stop using it, connect it another way if possible, avoid booting from it, pull only must-have files if reads are still stable, otherwise image it, then recover from the copy. And yeah, retire that disk after. It’s cooked.
