Assessing flaky HDDs

[Rigo]

G'day folks,
CrystalDiskInfo fails really bad drives that SMART didn't like. So when it passes a drive, I submit it to DiskGenius.
DiskGenius goes a little further and tries to go beyond SMART but even drives it passes might still have undetected badly flaky/bad sectors.
Anything else out there more reliable?

 

[mmerry]

People will probably hate me for this one, but you may consider SpinRite. I do not consider it a repair tool. I would only use it to "repair" the drive so I can get data off it and on to a new drive.

 

[HCHTech]

We see very few hard disks these days, but GSmartControl - any bad sectors = replace the drive....with an SSD. We use the Linux version on a dedicated machine setup for recovery, but there is a windows version as well.

 

[Larry Sabo]

I like Victoria for assessing SMART parameters, and doing a throughput test. If the drive has any reallocated sectors or sectors pending reallocation, it's a fail, regardless of an overall "Pass" score. There are a few other parameters that indicate that the drive is not to be trusted any more but I forget what they are, off hand. Hmmm, I just checked with ChatGPT and it gives a very helpful reply, including those other parameters I mentioned.

 

[britechguy]

Well, I'm the contrarian here when it comes to HDDs and need for replacement, because I've never encountered any HDD that's been in long term use that doesn't have a bad sector or two. It goes with the territory and they were designed to handle this eventuality gracefully. I always liked the following references, because they are accurate based upon my own experience over decades:
----
Bad Sectors Explained

Bad Sectors Explained: Why Hard Drives Get Bad Sectors and What You Can Do About It

“A few bad sectors don’t indicate that a hard drive is about to fail — they can just happen. However, if your hard drive is rapidly developing bad sectors, it may be a sign that your hard drive is failing.”

What Is a bad sector and how can I repair it?

Bad sectors are fairly common with normal computer use and the imperfections of the world we live in . . .
----

But, if the HDD is exhibiting "flaky behavior" it's already exhibiting clear signs of failure, and probably gathering bad sectors (and/or other things) at an increasing rate which is a huge red flag.

The moment any HDD starts exhibiting behavior I'd call "flaky" in any way, it's time to get the data off and replace it.

 

[Diggs]

Ddrescue running under Linux is my tool of choice (or DDRescue GUI) for failing/flakey hard drives. I understand there is now a Windows version but have never used it. If I can't get a good copy with ddrescue or it wants 2 weeks to copy then I send the drive off to the pros.

 

[timeshifter]

If I'm dealing with a PC or laptop and it doesn't already have an SSD I'm putting one in or recommending they replace the machine with a newer one. Only place I'm dealing with hard drives are on NAS these days. Sometimes for backup, but even for single PC backups I'm starting to use external SSD.

 

[britechguy]

Sometimes for backup, but even for single PC backups I'm starting to use external SSD.

Same here. SSDs have now become "cheap enough" that the price differential between the same capacity in SSD vs HDD is not insane.

I still prefer HDDs for long term "cold" storage, but that's not what most of my clients want/need these days. And now that the age of M365 is upon us coupled with Windows backup remembering apps and preferences, for those using both the need for backups on local media has become far less pressing.

 

[backwoodsman]

I mainly use Linux 'badblocks' to test hard drives. Assuming any data is backed up, either destructive or non-destructive surface scan, depending on whether I want to wipe it, or preserve what's on it. If it finds a few bad sectors that get relocated, that's normal and not a problem (drives relocate sectors only on a write request, not read, so usually won't relocate on a non-destructive scan). A bad drive will typically show up as a long list of bad sectors, which indicates a fatal problem of some kind. But if it passes the surface scan, and SMART doesn't show anything at or near EOL, it's good to go.

 

[YeOldeStonecat]

I gotta say, with around 35 years doing IT...for small biz mostly.....and with almost 3,000 computers under our watch in our RMM across our clients, we don't run drive test tools. Basic experience should tell you if a drive is starting to fail. Back in the days of old spinning platter drives..they degrade and wear out over time, their performance drops over time. If we suspected a drive is getting old and starting to fail, we simply clone to a new drive and move on. Spending time "testing" a drive, or "trying to repair a drive"...is simply not worth it. Would rather spend 30 seconds disconnecting the old drive and connecting it to the source side of our drive clone device, spending 60 seconds going into our shelf of new drives...grabbing one...connecting to the destination side of our drive clone device...spending 30 more seconds pushing the buttons on it to clone...and walk away and do other work until its done. Once completed, spend maybe another minute putting the new drive back into the customers computer and buttoning it up...booting up to check things. Total time spend...well under 5 minutes. OK..5 minutes if a laptop.

Each vendor usually has their own free tools you can download to thoroughly test their drives.

 

[labtech]

G'day folks,
CrystalDiskInfo fails really bad drives that SMART didn't like. So when it passes a drive, I submit it to DiskGenius.
DiskGenius goes a little further and tries to go beyond SMART but even drives it passes might still have undetected badly flaky/bad sectors.
Anything else out there more reliable?

What is your exact intent with those failed drives no matter how little or badly degraded they are?

 

[backwoodsman]

Spending time "testing" a drive, or "trying to repair a drive"...is simply not worth it.

True, of course. But it only takes a couple minutes to get badblocks started, and sometimes it's the quickest way to confirm the drive is bad. And most of the time when I do a surface scan, it's not because the drive might be failing, but because I need to wipe the drive and verify it's in good shape to use around the shop or resell to a budget-minded customer. Or maybe because the customer wants me to test the drive to confirm it's good while I have the machine anyway.

 

[YeOldeStonecat]

I guess for some markets. We won't risk putting a "used" drive in a customers computer, if it's suspect...it's replaced with new. The one we removed....generally tossed out, or..if it's newer (low hours)..we might keep in a junk parts drawer for some repurposing on a low priority old rig we deploy for...well, we don't really do that. So..yeah, NIST/DoD wipe and garbage for old drives.

 

[fincoder]

I've never encountered any HDD that's been in long term use that doesn't have a bad sector or two. It goes with the territory and they were designed to handle this eventuality gracefully.

That is true in theory, but in practice it's a bit different. From about the time when HDD prices plummeted, probably when SSDs first starting coming out and they were competing on price, HDD quality also plummeted. So for drives made in the last 10 years or so, any HDD found to have a bad sector should be swapped out. If not they will eventually develop more and start slowing down significantly before failing completely. Drives older than that 10-15 years would likely still be going strong with a couple of bad sectors.

But generally, HDDs are only tested fully if they're the boot drive and there are performance issues. So nowadays, I never fully test them (any PC running Windows 10/11 from HDD should be upgraded to SSD or binned). I just rely on Crystal Disk Info for general health of secondary data drives.

 

[Rigo]

We see very few hard disks these days, but GSmartControl - any bad sectors = replace the drive....with an SSD. We use the Linux version on a dedicated machine setup for recovery, but there is a windows version as well.

Trying it now to see if it see the problems on this drive.
I now know the drive has problems as RapidSpar skipped some sectors later extracted when cloning.
CrystalDiskInfo and DiskGenius passed the drive though the system was abnormally slow and some Windows tools couldn't even launch.

 

[Rigo]

Victoria

I'll give it a try as well on this patient

 

[Rigo]

Well, I'm the contrarian here when it comes to HDDs and need for replacement, because I've never encountered any HDD that's been in long term use that doesn't have a bad sector or two. It goes with the territory and they were designed to handle this eventuality gracefully. I always liked the following references, because they are accurate based upon my own experience over decades:
----
Bad Sectors Explained

Bad Sectors Explained: Why Hard Drives Get Bad Sectors and What You Can Do About It

“A few bad sectors don’t indicate that a hard drive is about to fail — they can just happen. However, if your hard drive is rapidly developing bad sectors, it may be a sign that your hard drive is failing.”

What Is a bad sector and how can I repair it?

Bad sectors are fairly common with normal computer use and the imperfections of the world we live in . . .
----

But, if the HDD is exhibiting "flaky behavior" it's already exhibiting clear signs of failure, and probably gathering bad sectors (and/or other things) at an increasing rate which is a huge red flag.

The moment any HDD starts exhibiting behavior I'd call "flaky" in any way, it's time to get the data off and replace it.

Yes I know that straight out of the box drives have bad sectors.
I remember the days when we had to launch a disk editor utility from floppy to configure the drives/systems before use.
Slow and unresponsive is not necessarily from dying drives so this needs to be confirmed before researching other potential sources/reasons.

 

[Rigo]

Ddrescue running under Linux is my tool of choice (or DDRescue GUI) for failing/flakey hard drives. I understand there is now a Windows version but have never used it. If I can't get a good copy with ddrescue or it wants 2 weeks to copy then I send the drive off to the pros.

It's more about confirming whether the drive needs to be replaced.
I favour Casper for drives in reasonable conditions.
When really bad I'd go for my RapidSpar imager.

 

[Rigo]

Spending time "testing" a drive, or "trying to repair a drive

Not everyone needs a blazing fast system (residendial/retired users) for upgrading to SSD if the original drive could still be used.
"Repairing" is definitely just a band-aid especially for drives that have exhausted spare sectors.

 

[Rigo]

What is your exact intent with those failed drives no matter how little or badly degraded they are?

From symptoms I might have suspicions that the drive is the problem but I need to confirm this before deciding/advising the next step.