The Warning Signs of Disk Failure: An Unexpected Tale of Gradual Decline and Urgent Replacement
Vložit
- čas přidán 26. 06. 2024
- ✴️ Of course it would be the largest disk in my system.
✴️ Slow failures
Hardware failure can happen suddenly or gradually, often making diagnosis a challenge. Gradual failure of a hard disk can manifest as system slowdowns, increased errors, blue screens, and slow operations during heavy disk usage. Replacing a failing drive can restore both performance and stability. Above all, though, it’s another reminder to always be backed up.
Updates, related links, and more discussion: askleo.com/170689
🔔 Subscribe to the Ask Leo! CZcams channel for more tech videos & answers: go.askleo.com/ytsub
✅ Watch next ▶ It Was My Time to Reformat and Reinstall ▶ • It Was My Time to Refo...
Chapters
0:00 Warning Signs of Disk Failure
0:40 Gradual slowdown
1:20 Post install
2:40 Things got worse
4:00 Replacement
6:00 Dropbox and OneDrive return
7:00 Still to come
❤️ My best articles: go.askleo.com/best
❤️ My Most Important Article: go.askleo.com/number1
More Ask Leo!
☑️ askleo.com to get your questions answered
☑️ newsletter.askleo.com to subscribe to the Confident Computing newsletter.
☑️ askleo.com/patron to help support Ask Leo!
☑️ askleo.com/all-the-different-... for even more!
#askleo #disk #failure - Věda a technologie
✅ Watch next ▶ It Was My Time to Reformat and Reinstall ▶ czcams.com/video/lNZOsvzjiZQ/video.html
I would consider plugging the replacement into the SATA adapter and copying off the files *then* doing the swap. I always feel there is a risk that lumping a drive about, potentially with a physical or electrical issue, could result in a completely failed source drive so get the data off it *before* handling it. I've never had a complete drive failure like this but it makes me feel better!
I thumbed you, good point. However, IME not really an issue. I've pulled a lot of drives out and connected them externally via a USB or other connector to copy the data into a system.
If you're being careful you're not going to drop the drive. I'd rather have the computer in it's end-state (physically) when the (long time spent) copying is complete.
Also on the topic of "dropping a drive". They're built to take a rather hard multiple-G shock. So if the drive has spun down, you'd have to drop it several feet or more to cause harm. If you drop it 6 inches (18 cm) it's not likely to be harmed.
I would also recommend you have a separate computer (whose system-state doesn't matter) to perform your DOD erasure. It would be rather unfortunate to accidentally erase the system you're actually trying to fix.
The only other consideration for recovery I think is that the computer's internal controller may be more capable of data retrieval than an external one on a USB connector (more commands it can issue, and what-have-you). But probably not. IDE for example, has most of the drive controller tech on the actual drive, presumably SATA and NVME, etc also do.
It's one of those things, doesn't really matter internal or external. Personal preference.
In this case, I would have just done the restores from the Internet again. Sure it would take a few more hours, but I don't know the quality of the data coming off of the old drive. So I'd rather just get it off the Internet once and not have to be concerned with it. Especially for a client who's paying me to do work.
Anyone who's ever copied data off of a failing drive could tell you, it's fraught with random failures. I don't know how robust the online storage systems are or if a bad file appearing on my system is going to get copied out and over-written to the Internet service.
As per the failures, Leo covered that with a couple command switches. But still, how robust is that for the online services and updating files? Are they all equally capable? (probably not)
As someone noted... Checking the SYSTEM EVENT LOG (before the Windows RESET/REINSTALL) would almost certainly have shown a bunch of I/O errors on that hard drive. Once you find the I/O failure, you can sort on it to get an idea of how frequent they have become... when they first started, etc. But of course the reset/reinstall would have wiped out the event log. Regardless, new error events WOULD have appeared as the problems reoccurred after the reinstall.
As for the root cause of the disk failure being a loose connector... Yeah this is possible. I have seen SATA connections go intermittent after LONG periods of time. Usually something to do with contaminants left on the contacts and exposure to moving air. Disconnecting the cable and using compressed air followed by a quick spray of contact cleaner on both sides will generally resolve that for good. If there actual internal disk errors, the SMART data monitoring in the drive should have indicated that. Once the SMART data is showing an error count, then you probably want to replace the drive. Sure, it can remap failed sectors (causing slowdowns), but do you really want to be counting on a disk that is starting to remap sectors...? Predictive failure - "there's your sign".
However if the SMART data is "clean", then it probably was a SATA or power supply issue.
LOADING... *CLICK, CLICK, CLICK* LOADING... *CLICK, CLICK...* Crashing applications, poor performance. BSODs."Cannot read file." The days of crashing system hard drives. Fun times!
In the old days MFM and rll drives did make a lot of noise when they started going bad.
@@davidmartin8211 huh I never noticed.. I only had 4 or 5 of them though. Oh and none of them went bad on me (knock on full-height drive). The days of the Gigantic One-Hundred-Twenty (megabyte) hard drives!! wow! The drive was gigantic, really!! Not kidding!
Floppy discs now.. Enn ehnn eeehhhnnn en en en en en en ehhhhhhnnn en en eehhhnnn...
Oh and of course, a wise person once said "On a blank disc, you can seek forever..." oh I better stop making jokes now. they might start causing skips...
1. Offload files.
2. Install new drive, restore files.
2. Use WD software to zero wipe.
3. If you have a spare pc that does not use UEFI, run Spinrite. On that size drive, it would take days. Maybe level 3. Level 4 might repair it.
4. Spinrite is just a way to try and salvage the hardware. It will retain the data if you leave it on the drive, but why take chances. The current version boots on a CD or thumb drive you create with Spinrite, and does not allow UEFI. A new version is promised to allow that. Disclaimer - I am not affiliated with Spinrite, and have not used it recently.
Also note that if the disk is in fact bad, the time can skyrocket. I think that it has the ability to forgo recovering existing data. This would speed things up a lot.
I would rather buy a new drive than paying $89 for spinrite only to find that you need to buy a new drive.
@@Klaus2m5 Reasonable. When I bought it, hard drives were very expensive.
If I were to do this, I would base it on the age of the drive. If it's 4+ years, get rid of the drive, it's likely at end-of-life. If it's a few months, again it might be end of life but there's a good chance you'll recover it. If you're a lone hobbiest like probably most people here -- don't bother, just get a new drive as @Klaus2m5 suggests.
They've done a few studies including a very large one by Google that found the drives have a tendency to either fail very early in life or very late. If they last beyond a certain time (I think it was 5 or 6 months) they're statistically unlikely to die until they've been very thoroughly used, out to 5 years or more. By then you can likely get a faster, larger, or some other attribute drive for the same or cheaper price. Note -- they do not get "more predictably reliable as the industry ages", especially now as we have SSD replacements coming.
We haven't quite flipped to SSD being cheaper than discs, but it's getting closer. They may be holding back on this line deliberately to soak us, I don't mind the prices are reasonable, IMHO.
Love your content. I have been following you for many years.
If you have a back up image of the failing hard drive, could you have restored that image onto the new hard drive instead of the copying process that you did?
Yes.
Yes sir. As with any kind of backup, you're copying data around, that's all we're talking about when talking about backups.
Well, and also ensuring accuracy, but I was taking that and other issues for granted for the sake of summarization.
In Leo's specific case though, you're talking about multiple Terabytes of data. Where would you store a multiple Terabyte image file? It would have to be another multi-terabyte receptacle (presumably a drive, but could be tape or WORM/WARM optical disc, NAS system, SAN system...).
I'm normally not a fan of online storage. I don't like the idea of private information stored in a place where someone else can potentially access it. However if the data is sufficiently low-impact on you personally then doing what Leo is doing is a great idea.
off-topic factoid -- The rumored AGI supposedly requires an entire google compute facility. How many years before the AI software becomes efficient enough and the hardware becomes fast enough for you to have an AGI in your own PC/cell/chip-in-your-head? Or will they ever allow AGI to the masses? Interesting questions.
Booted up a cold storage drive over weekend. Got SMART error. I didn't expect a helium drive to leak...Still works so I can get data off.
Seems that sometimes HDDs fail without SMART warning. I used HDDScan and found some block reads taking > 500ms. Still SMART was not counting any uncorrectables. I could have tried a low level format but choose to replace the drive with an SSD anyway.
Just last week, My sign was when the drive enclosure started smoking with the smell of burning plastic. This was caused by a defective power brick or supply. Zapped the NTFS partition.
No more moving parts - less headaches.
I am not an IT person but I have learned a few things already. I guess your failing HD was the HDD, not SSD, which is said it just jumps off the cliff, not giving regular warnings.
I wanted to ask: what happened with the testing software like crystaldiskinfo, or HDDscan? I used both to check my drive after the manufacturer's system gave me a warning message that the hard drive is failing and needs to be replaced. I scanned it with both software and all came out good. Only on deeper scan via HDDscan, the graph of analysis was up and down, indicating there might be an issue. It is a company's machine, I'll let them deal with that.
Spinrite has a scan option. Note that it disables internal retries. If the sector is even a little bit bad, it will say so.
Two items:
1) Salvaging your old drive:
The old drive might have had bad sectors, which is a bad sign.
After having a backup of that drive's data, I would have performed a long format (the type that takes ½ of forever to complete). It makes no assumptions that the drive is good. Instead, it checks every sector during the formatting operation, and will mark each bad sector such that Windows will not see those parts of your drive's platters. It would be as if they do not exist, leaving only good sectors for Windows to see.
Of course, that still might be a temporary fix, if the drive is failing, as more sectors could go bad (or more bits within sectors go bad) after the formatting is done.
Drives will function properly if X number of bits in a sector have gone bad (where X is probably 10 or less). If enough bits have gone bad, but the sector can still ultimately work, then that could cause slower operational times. And I do not think that the drive will take preventive measures to deal with the rising number of bad bits in a sector (and at some point, you will have data loss).
Steve Gibson, of Gibson Research Corporation (GRC) has a SpinRite tool that monitors the health of hard drives. It works proactively. When it detects that X number of bits in a sector have an issue, it will copy that sector's data to a problem-free sector, and then mark the original sector as bad.
Short of a catastrophic drive failure, running SpinRite on a regular basis (perhaps monthly, or quarterly) will continually mark problem areas, keeping your drive running as if nothing is wrong.
2) An alternative to "xcopy".
@4:44 -- I recommend "robocopy".
robocopy f:\ e:\ /mir
The above translates to "f:\" being the source directory, "e:\" being the destination directory, and "/mir" stands for mirror, which means it will copy anything and everything.
robocopy will show you its progress for each file that it is copying. So if your copying job seems like it might have stalled, then if you are using robocopy, you will know (no guessing). It reports in 1 tenth of 1% increments.
Running "robocopy/?" will list a myriad of options, to handle just about any special copying criteria you will have.
Get to know robocopy. Make it your friend. You can, of course, copy-paste or drag folders via the GUI, or use xcopy (nothing wrong with those choices). But when you need to do something complicated, robocopy will have an option to handle your needs. And the /mir option makes it simple to make a mirror copy of folder "A" to folder "B". Note, however, that every copying tool will have issues with files that are in use. So robocopy cannot make a mirror image of your "C:" drive.
The best use of HDD is backup these days.
Besides bad sectors, there are weak sectors too. They can be fixable, but not for a long time. That's where you have been. Do NOT trust this drive any more, no matter what chkdsk /R may result too. It cannot detect delays and weak sectors, which in turn become bad.
Had a Lenovo Ideapad, running Linux, start getting slower and slower. I finally replaced the spinning rust disk with a new SSD. Fixed the issue immediately. I copied the spinner over to the SSD first, and didn't open things up until I had a good image on the new drive. The old spinner now lives in an external drive case, and seems to work,but, whatever its problem was when inside the laptop,it is disqualified from doing anything important for me.
When my hd failed it wasn’t clear if it was the drive that failed or the hard disk controller that was the culprit. Given that the computer itself was getting long in the tooth it seemed prudent to replace it as it was certainly beyond my understanding to diagnose the issue and repairing it at a shop was not fiscally prudent
xcopy also has verify option
I use Teracopy.
Lots of great information. Very rare do I format my drive and start from a new Windows. If I do instead of formatting the current drive, I would get a new drive to use and just unplug the old drive as a backup in case the Windows setup have issues. Would you not use CrystalInfo to get the status of the drive?
I have experienced a computer completely freeze for a minute or two on multiple occasions before
I never thought much of it until one day, just after i had installed an SSD to copy my data onto (i think i was planning on upgrading the system to SSDs in multiple stages)
And just as i started to copy the files to the SSD, the system froze like it had done many times before, but this time it didn’t recover
Tried rebooting the system, instead of launching Windows, it opened the firmware settings
The HDD seemed to be dead (notice i said SEEMED to be, as what was actually going on was extremely weird, but I didn’t know that at the time)
Most of my failed drives have been dropped or external drives. Ps these days do NOT use Amazon, its full of used items as new, and counterfeits. Use a dedicated trusted computer outlet instead, no third parties
Leo's description is a classic case of a slowly degrading drive. It is best to check the drive with a SMART tool first.
Unfortunately, if this is your main drive and the operating system is occupying portions of the drive that are damaged then any amount of reinstallation are essentially useless. This, sadly, is a Stark warning and wake up call for anyone who purchases a computer with play hard drive glued onto the motherboard!!
Liked and subbed 🏆
yes happened to me in my lenovo ideapad laptop actually my toshiba 1tb mecanical drive died on me ran crystal diskinfo and it said caution had to replace it with 2.5 icoolax ssd it lasted 3 years and my ssd is still going great luckley i have everything backup on external hdd btw did you run crystaldisk info first on it lemmi know in my case that did not work had to many bad sectors when tech guy wanted to clone it with acronus it failed cause of bad sectors
2,5 HDDs are SMR nowadays cuz 7 mm forced thickness so they could really go back to CMR cuz then it would have some niche . 3-4-5 TB 2,5 drives in CMR would be nice
Never knew there was a command for this. Have always used disc cloning software.
Before there was cloning software, there was DOS. I would assume most people today do not have a copy of DOS sitting around.
@@ronrollins381 I have DOS Box to play some 80's computer games. Haha.
@@ronrollins381 I still have DOS 5.0 & 6.22
Does the order of the options ("/D/E/C/H") matter?
No.
@@askleonotenboom, I thought so. Thanks for the confirmation.
Xcopy? Why not robocopy? I have found robocopy to be a real lifesaver.
For the most part copying is for the most part copying. I use both, but xcopy's options are simpler.
@@askleonotenboom
robocopy f:\ e:\ /mir (as in mirror)
Have you used Teracopy?
@@colt5189 I never heard of Teracopy. Apparently, it has been around for several years.
From my brief research, it looks like a GUI alternative to Explorer, when copying files. You still use Explorer to copy / paste or to identify the files in question. But then Teracopy takes over to perform the actual copying.
Teracopy appears to identify when you are copying numerous small files, and will kick off concurrent / simultaneous copying jobs, as long as SSDs are involved.
Explorer, on the other hand, will copy each of those small files, 1 by 1. So Teracopy will be much faster.
But there is no comparison with robocopy, when it comes to command-line options, to customize your copy request.
And with robocopy, it is easy to repeat your copy job, or put it in a bat file.
Senior : any backup copy without verification is not a backup 😂
When I copy/paste, I use Teracopy that uses checksums to verify the pasted copy.
Heavily used HD, 3.5 years old and does an uninstall, format and reinstall on old dog hardware.
Then finds out not worth it and starts the process over again with a new HD installed. Uncle Leo slipping here?
Uh, no. The drive was not my system drive, and it was replaced before this started.
Cdi takes three seconds to run
A smart person would have checked SMART info in the drive, plus check Event Monitor for the specific errors.
Guessing the problem was a loose power connector, so it should work just fine once you install it somewhere else.
Chkdsk will not fix this. A “soft” HD error would not cause a blue screen.
If the HD seems to work ok now, it was a loose cable.
Putting stuff on 8 TB drive? Think about eggs. Put all your eggs in one basket or have several smaller baskets and be safer if there is a hard crash! Playing games to save a few bucks indicates your data is a bunch of junk that you probably could throw away anyway!
According to the video, the data was backed up by cloud storage, and I’m sure there were other backups as well.
I’m also sure RAID would have been used if he required high availability.
A suggested idea is to create multiple smaller partitions for example 2 GB. a bad area of the drive could scramble one partition and leave the other ones active. Fyi. drives are appearing in the 20 plus gigabyte rage.
It was one of multiple copies of the data, so backups were in place. No data would have been lost had the drive died completely.
my 1tb 2.5 inch toshiba hdd has zero health but i am using it for moving files. the speed has got a little slow. i have been using it from 2013 :) ,,,, old but my work could be done on it. i dont know how long will it run .... may be 10 to 20 years:)