Windows won't check hard drive

I’m running 64-bit Windows 7. Last time I rebooted, the OS decided that it needed to check the integrity of my C drive, a SATA drive. The notification screen popped up and started counting down, giving me the opportunity to hit a key to abort the process.

I let it start and wandered off to other things. An hour later the screen was still stuck at “press a key in (1) second”. The hard drive light was flashing briefly and regularly, so something was happening. Taking a chance, I hard-rebooted the system and tried again. Same result after about 45 minutes.

When I abort the check, Windows boots normally and nothing seems affected. Is this a sign of Very Bad Things about to happen to the drive? If not, how can I remove whatever flag is telling Windows to check the drive on each boot? What would trigger this flag? I haven’t had any bluescreens or power outages recently.

Thanks!

SpinRite time. Sounds like a damaged block right where chkdsk wants to start reading.

I ran SpinRite’s faster setting (level 2) and it found nothing. When I rebooted into Windows, however, the requested chkdsk kicked off and executed smoothly, no problem detected there either.

So…SpinRite fixed it? I think…either way, it works now. I’ll run SpinRite’s more intensive scan (level 4) overnight and see what develops.

Thanks Christoph!

GODDAMNIT.

Don’t run spinrite until you’ve run a SMART test to see IF bad sectors are the fucking problem. Seriously people. Spinright is a REPAIR tool, NOT a diagnostic tool. /rage

Well, it worked! So it functioned as a repair tool. Things didn’t work right before, tool ran, things work after.

So there. Correlation is casuation. QED.

That’s completely wrong. In my experience SMART is worthless, it never shows any errors as long as the hard disk is accessible at all. And Spinrite is most definitely a diagnostic tool, it finds precisely those defective sectors that SMART never detects. There’s a read-only scan, too, in case you don’t want to risk writing to a disk.

Perhaps some head alignment issue? Anyway, good to hear that it works now but you should make sure to have a recent backup handy, and maybe think of a replacement disk. Such errors often come back and multiply. How old is your disk? My desktop drives generally started showing defects after about 2 years of use.

It’s 4 years old. I’m backing up to an external drive now. Given that it’s my OS disk, this is the perfect opportunity to swap in an SSD drive.

What software do you use? Do you actually check the pending/bad sector raw data or just look at the pretty numbers?

I’ve used SMART info to replace drives that were showing signs of wear on numerous occasions. I cannot fathom how it’s more productive to run Spinrite for 10 hours rather than taking 2 minutes to check SMART info and find out if there are bad/pending sectors.

I’m looking at the SMART data shown by any utility that shows such data, such as Spinrite itself or Speccy, and also the WD disk check utility when I was using WD drives. Routinely, the read error rate was 100 followed by a row of zeroes for the raw data, which I was assuming means SMART has detected no errors. Is that incorrect?

Spinrite takes 10 hours only when you do a full surface refresh, by the way. A simple read check on all sectors takes maybe 10-30 minutes. And then you know which sector is bad, not just that there’s a bad sector (which you already know anyway). Or are you saying you have a utility that actually extracts this data from SMART?

Yeah, unfortunately, that data is often useless. If you run Speedfan and check the SMART data, there’s an option to “view detailed” - Which kicks you to an awesome web page that tells you the actual useful stuff. I’ll detail some examples so you’ll get an idea of what this sort of stuff has helped me deal with.

Case 1

This report was from a system that suddenly developed errors trying to boot into Windows. The client had attempted Windows 7 startup repair, which was worthless as usual. First thing I did was slave the drive to the diagnostic machine and run Speedfan, which got me this report.

2 pending sectors. A 1-hour scan of HDDRegenerator fixed those, repaired those sectors, found no bad ones, and the system was restored to functionality.

Case 2

This system had experienced a virus infection and I instructed the client to shut it down. When I got to the system, I was able to boot into Windows and I began running diagnostics. The system bluescreened at that point, indicated a code that often referred to drive read errors, so I immediately stopped and slaved the drive.

Tons of bad, pending, and reallocated sectors. Mind you, up until when I started working on the system, they were using the computer despite the infection. I ran HDDRegenerator to repair the pending sectors, and I was able to clone the drive with Acronis without any errors. I then did the virus removal on the new drive and go the system back up and running.

Case 3

Reformatting computer and just doing due diligence uncovered this system had 11 reallocated sectors. That’s not a huge worry, but it is something to pay attention to. One month later, there’s 14 reallocated. I’ve recommended to this client to budget for a hard drive replacement in the next 60 days, just to be safe.

Case 4

This is probably my favorite to date, as I’ve had this situation occur only a couple times and there’s a few unique parts to it. First off, this hard drive was showing no problems. The user noticed no performance issues, Windows was running fine, the system never crashed. Upon boot there was no warnings about the drive being bad, as sometimes happens with failing/dead drives.

But it’s standard procedure for me that if I touch a computer, I install Speedfan and run SMART tests. And sure enough, I find this thing has over 8000 reallocated sectors, and nearly 2000 pending sectors. How on earth it was still running, I don’t know. Even more miraculously, I was able to clone the drive without a single error. The drive was stable enough to remap the bad sectors without data loss, but obviously wasn’t stable enough to not be racking up a huge number of these sectors. And somehow, none of those 1800 pending sectors fell on an area of the drive that contained system data.

Interestingly enough this is also the first duplicate drive I’ve seen in my failure database. A little over a year ago, I started keeping track of the details on all failed hard drives. I did a data recovery on this exact Samsung model in another client’s system around the time I started that database. Both were shipped from Dell in 2009, fairly new systems by all accounts, but clearly Samsung drives are just as crappy as I’ve always suspected.

In almost all of these cases, I can imagine running CHKDSK would’ve exacerbated recovery efforts, and running Spinrite would’ve just wasted valuable time. But that’s my opinion, I haven’t had a scenario that was remedied by Spinrite in over 3 years.

</end wall of text>

Thanks for the detailed examples, that’s quite interesting. I think the error-related entries other than read error count showed default values on my failing hard disks, too, but I can’t honestly be sure from memory. I never thought to run SpeedFan’s detailed analysis so it’s quite possible I missed something there. On the other hand, I try to repair or replace disks at the first sign of error, unlike the rather extreme cases with some of your customers, so perhaps SMART simply had not gotten around to record the error data.

Either way, Spinrite’s scan is ultimately another way to arrive at bad sector data, and its disk refreshing function should work the same way as HD Regenerator which you use, so it ought to work out the same in the end.

Yeah, not remembering specifics is why I started keeping detailed logs. It’s really good to be able to know “Hey, this client had a perfect SMART scan 6 months ago, but now they’ve got 22 reallocated sectors.”

I don’t know if you were looking at the reallocated/pending/bad counts, but you mention “read error count” and that’s one of those metrics that I find to be worthless in this day and age where read errors are a fact of hard drive operation, at least taken in isolation.

Check out Case 3 and Case 3 part 2. See that read error count? Do you notice it DECREASED from 1 to 2? So did hardware ECC recovered. How has the read error count gone down? What the hell does that mean? Heck, even the Seek Error Rate has somehow improved! The only useful piece of data in there is that there are 3 more reallocated sectors than there were a month ago.

Look at Case4 - There’s no problem with the read error rate, there’s no “reallocated” sectors according the drive - The only sliders that are in “bad” states are the “duration” ones, which don’t tell anything about the health of a drive. And yet, tons of pending and offline sectors.

Interestingly, neither the pending or offline sectors counts ever get high enough to make that health slider change position. Which is another oddity that makes Case 4 so weird, despite the fact there are pending and offline sectors out the wazoo, the RAW read error rate is practically nil and the health of the drive, by all other metrics, is 100%. Yet it’s pretty obvious that Case 4 needs an immediate drive replacement, and should have flagged a warning by SMART during the BIOS POST.

I think this is key as to why people think SMART is worthless - When you’re running a PASS/FAIL test, it’s clearly not basing that on all reported drive metrics.

Anyways, you label these cases as extreme, I don’t consider them so. This is snapshot of what I’ve run into in the last 8 weeks. I’d call these scenarios pretty typical. Since I’m running Speedfan on ever drive I touch, I’m catching a lot of stuff preemptively. Extreme was the hard drive that had 30,000 bad sectors and I was still able to recover 99% of the user’s data.