goHardDrive via eBay has
12TB HGST Ultrastar HC520 7.2K RPM 6Gb/s SATA 3.5" Internal Hard Drive (Certified Refurbished, HUH721212ALE601)
+ 5-Year Seller Warranty on sale for
$79.99.
Shipping is free.
Thanks to Community Member
xrossastrike for finding this deal
- Note: Includes a 1-Year Allstate Warranty + a 5-Year Warranty from the Reseller.
Specs:
- 12TB Internal Capacity
- 3.5" Form Factor
- SATA 6Gb/s Interface
- 256MB Cache
- 7.2K RPM Spindle Speed
161 Comments
Your comment cannot be blank.
Featured Comments
6 of one, a half dozen of the other.
If you're doing drive imaging backups of something like a system drive under Windows, I'd recommend Macrium Reflect.
If you're just doing a basic sync, I like FreeFileSync personally.
If you can, I like to buy slightly different models or try to buy from a different vendor to differentiate model numbers or at least HDD batch numbers. Just in case a firmware bug or some other unexpected flaw for that particular model or batch eats one of my drives. At least the other one (or other set) will likely survive. Just a thought.
Also don't forget to do a full surface test, which will likely take more than a day at these drive sizes. You can use something as simple as WD Data Lifeguard for this.
Sign up for a Slickdeals account to remove this ad.
It's much more like buying a used car than old food. Can you buy a used car and get 10 or more years out of it, sure. Can you get a lemon that dies within a year, also sure. Does a new car guarantee a problem-free vehicle, absolutely not.
The price difference needs to be enormous for most people to bite on used vs new, and it is here. So is it worth a shot, IMO Yes because you should have backups anyway. If the seller stands behind their hardware, what exactly is the risk? No, really, tell me.
Because a new drive can also fail, so if you're relying on drives not to fail as your data integrity strategy..... that's ridiculous. Might as well buy drives for less and take your chances. Plus, you should be aggressively testing ANY new (to you) drive, regardless of its age or where it came from. Transit introduces all kinds of issues.
Link 1 [youtube.com]
Link 2 [imgur.com]
Link 1 [youtube.com]
Link 2 [imgur.com]
We need online therapists...
Sign up for a Slickdeals account to remove this ad.
The default setting in the Victoria program is 10,000 which is on the very high side.
A very low ms number setting may erroneously show a drive full of slow/weak sectors, and a very high number will show no such slow/weak sectors even though they may exist.
To do a full read test on these 10Tb and 12Tb drives will take about a week using Victoria.
It may show something like this in red letters for any weak/slow sectors that were not picked up by the Windows CHKDSK test:
9:59:07 : Warning! Block start at 4579686400 (2.3 TB) = 9203 ms
10:37:42 : Warning! Block start at 4698906624 (2.4 TB) = 8532 ms
This 10Tb hard drive is still running the Victoria test after 3 days with about 50% done.
The above 2 errors showed up on this 10Tb drive that had the chkdsk test run successfully with 0 bad sectors:
Stage 5: Looking for bad, free clusters ...
2441504550 free clusters processed.
Free space verification is complete.
Phase duration (Free space recovery): 15.09 hours.
Windows has scanned the file system and found no problems.
No further action is required.
9537517 MB total disk space.
35904 KB in 6 files.
72 KB in 13 indexes.
0 KB in bad sectors.
364251 KB in use by the system.
65536 KB occupied by the log file.
9537127 MB available on disk.
4096 bytes in each allocation unit.
2441604607 total allocation units on disk.
2441504551 allocation units available on disk.
Total duration: 15.09 hours (54356324 ms).
This is some relevant info on a 7200 rpm drive's ms timeout:
https://superuser.com/questions/9...ading-time
It seems that a hard drive with a high ms timeout sectors should be corrected to mark those sectors as bad, because they'll go bad soon.
An optimal sector timeout for a 7200 rpm hard drive would be about 8.33 ms or at worst perhaps the 250 ms suggested by the other HDD site.
It definitely should not be in the multi thousand ms, and if there are such slow /weak sectors they should be marked as bad and excluded from use.
IF there are hundreds or thousands such slow/weak sectors found, IF these tests are run properly, it's probably best to return these drives.
Another thing that was pointed online was to do these tests plugged in to the SATA interface and not use a USB external box.
It's a good idea to run the Windows chkdsk, and these more sophisticated tests on these used drives, BUT, one MUST know how to run them reliably, and that takes some research online.
These tests should not be run if one doesn't know how to use them reliably. Just stick with the standard Windows CHKDSK test with the proper switches.
Anyone with more info on this chime in please.
Such HDD integrity tests should probably run in any drive, new or used, but again one must know how to run them reliably.
The default setting in the Victoria program is 10,000 which is on the very high side.
A very low ms number setting may erroneously show a drive full of slow/weak sectors, and a very high number will show no such slow/weak sectors even though they may exist.
To do a full read test on these 10Tb and 12Tb drives will take about a week using Victoria.
It may show something like this in red letters for any weak/slow sectors that were not picked up by the Windows CHKDSK test:
9:59:07 : Warning! Block start at 4579686400 (2.3 TB) = 9203 ms
10:37:42 : Warning! Block start at 4698906624 (2.4 TB) = 8532 ms
This 10Tb hard drive is still running the Victoria test after 3 days with about 50% done.
This is some relevant info on a 7200 rpm drive's ms timeout:
https://superuser.com/questions/9...ading-time
It seems that a hard drive with a high ms timeout sectors should be corrected to mark those sectors as bad, because they'll go bad soon.
An optimal sector timeout for a 7200 rpm hard drive would be about 8.33 ms or at worst perhaps the 250 ms suggested by the other HDD site.
It definitely should not be in the multi thousand ms, and if there are such slow /weak sectors they should be marked as bad and excluded from use.
IF there are hundreds or thousands such slow/weak sectors found, IF these tests are run properly, it's probably best to return these drives.
Another thing that was pointed online was to do these tests plugged in to the SATA interface and not use a USB external box.
It's a good idea to run the Windows chkdsk, and these more sophisticated tests on these used drives, BUT, one MUST know how to run them reliably, and that takes some research online.
These tests should not be run if one doesn't know how to use them reliably. Just stick with the standard Windows CHKDSK test with the proper switches.
Anyone with more info on this chime in please.
Such HDD integrity tests should probably run in any drive, new or used, but again one must know how to run them reliably.
I disagree with your comment that one should not run these tests unless they fully understand what they're doing. The standard read tests are non-destructive. The biggest problem is people misinterpreting the results, which can happen with basically any of these HD utilities.
For the purposes of testing a drive, I would not mess with the default timeout. If that setting does what I think it does (I can't find any English documentation), you generally wouldn't want to set it as low as 250ms. Yes, I think I know where you got that number but it was in regards to a totally different situation.
I'm not sure if it can be adjusted but I believe Victoria will normally log any blocks in the "orange" or worse range (which appears to be >1 second). HDDScan can be adjusted and defaults to reporting sectors >50ms. I think HDDScan's default is too low and Victoria's is a bit high. I prefer somewhere around 150-250ms. 8.33ms would absolutely not be an acceptable timeout. A timeout is a maximum. 8.33ms would be nearly optimal. Even very fast hard drives have random access times of roughly 10-20ms.
There are a few more factors that serve to muddy the water further. One is the block size. I'm not sure how these programs determine the default block size to test (probably based on the drive capacity) but it often does not align with the drive's physical or logical sector size. On a 12TB drive, Victoria selects a 2KB block size and HDDScan picks 1KB. The drives have 4K physical sectors and 512 byte logical sectors. It might be a good idea to bump the block size to 4KB, so they align with the drive's physical sector boundaries.
Now, here's a really important warning. Both of these programs run on Windows. Windows itself and other programs may attempt to access the drive, while the scan is running. This can often cause false positives for slow sectors. It can be hard to avoid this on a regular Windows installation. For this reason, I prefer to run these utilities from a live Windows PE session (using Hiren's BootCD PE). Unfortunately, this gets into areas too complicated for many people. My best advice for the average user is, if one of these programs logs slow sectors, run a couple additional scans and see if the same sectors come up slow. It doesn't have to be a whole drive scan, you can set the programs to check only portions of the drive. You'd want to start the rescan at least a little ways before the sectors in question.
If these tests are performed in a way that produces accurate results, you don't want to see ANY bad or weak/slow (say >250ms) sectors. The eBay ad for the drive in this deal explicitly highlights a drive with "perfect" surface scan results. In general, if a drive had even a hundred weak sectors, I'd classify it as failing. I don't think I've seen a drive with >20 bad/weak sectors where the damage wasn't spreading. Anything more than 0 is cause for concern.
BTW, the link you posted isn't really relevant to this discussion. It's a question based on a purely theoretical, ideal, and incomplete situation.
Have you bought a number of GOHARDRIVES 12tb drives , like in this thread ... and how many failed the tests that you performed on them. Was it like 20 percent failed. In my last shipment a few days ago 2 drives didn't INITALISE and of the 2 drives that Initalised ... one drive in SMART showed YELLOW CAUTION on it, and only 1 drive passed Smart and using HHD had 0 bad sectors at default settings on surface read mode (took about 14 hours).
What is your experience on these drives? Sounds like you have knowledge most of us don't have with these drives.
I disagree with your comment that one should not run these tests unless they fully understand what they're doing. The standard read tests are non-destructive. The biggest problem is people misinterpreting the results, which can happen with basically any of these HD utilities.
For the purposes of testing a drive, I would not mess with the default timeout. If that setting does what I think it does (I can't find any English documentation), you generally wouldn't want to set it as low as 250ms. Yes, I think I know where you got that number but it was in regards to a totally different situation.
I'm not sure if it can be adjusted but I believe Victoria will normally log any blocks in the "orange" or worse range (which appears to be >1 second). HDDScan can be adjusted and defaults to reporting sectors >50ms. I think HDDScan's default is too low and Victoria's is a bit high. I prefer somewhere around 150-250ms. 8.33ms would absolutely not be an acceptable timeout. A timeout is a maximum. 8.33ms would be nearly optimal. Even very fast hard drives have random access times of roughly 10-20ms.
There are a few more factors that serve to muddy the water further. One is the block size. I'm not sure how these programs determine the default block size to test (probably based on the drive capacity) but it often does not align with the drive's physical or logical sector size. On a 12TB drive, Victoria selects a 2KB block size and HDDScan picks 1KB. The drives have 4K physical sectors and 512 byte logical sectors. It might be a good idea to bump the block size to 4KB, so they align with the drive's physical sector boundaries.
Now, here's a really important warning. Both of these programs run on Windows. Windows itself and other programs may attempt to access the drive, while the scan is running. This can often cause false positives for slow sectors. It can be hard to avoid this on a regular Windows installation. For this reason, I prefer to run these utilities from a live Windows PE session (using Hiren's BootCD PE). Unfortunately, this gets into areas too complicated for many people. My best advice for the average user is, if one of these programs logs slow sectors, run a couple additional scans and see if the same sectors come up slow. It doesn't have to be a whole drive scan, you can set the programs to check only portions of the drive. You'd want to start the rescan at least a little ways before the sectors in question.
If these tests are performed in a way that produces accurate results, you don't want to see ANY bad or weak/slow (say >250ms) sectors. The eBay ad for the drive in this deal explicitly highlights a drive with "perfect" surface scan results. In general, if a drive had even a hundred weak sectors, I'd classify it as failing. I don't think I've seen a drive with >20 bad/weak sectors where the damage wasn't spreading. Anything more than 0 is cause for concern.
BTW, the link you posted isn't really relevant to this discussion. It's a question based on a purely theoretical, ideal, and incomplete situation.
For the Victoria HD test I left everything at default. That's why I mentioned the 10,000 ms timeout.
I agree with you that even though the timeout is set that high there must be another criterion when this test is run since it picked up ms timeouts less than the 10,000 ms default setting at 9203 and 8532 ms.
Also, you must be right about Windows trying to access the drive causing erroneous slow/weak sectors. That may be the case here with these two high ms timeout results.
Also, I'm running these tests on a USB external HD enclosure, since I have only a couple of laptops, no SATA interface as these tests should be run, and that may affect this Victoria test in taking so long and perhaps providing erroneous results.
The CHKDSK tests that I did on both of these 10Tb drives took about 16 hours and both came back with NO bad sectors/clusters.
Both drives were formatted with the default NTFS 4K allocation unit size or block size before the CHKDSK and the Victoria tests were run.
The 8.33 ms timeout link I mentioned is of course ideal, and it was mentioned purely as relevant information.
I haven't run the HDDScan test yet.
Screenshots of the Victoria test currently running are posted.
You are right about Windows accessing the drive is causing erroneous slow/weak sectors.
The screenshot below with the third slow/weak sector was when the drive was accessed by Windows to see the SMART of the drive.
So, most likely the other two listed weak/slow sectors are erroneous too.
Have you bought a number of GOHARDRIVES 12tb drives , like in this thread ... and how many failed the tests that you performed on them. Was it like 20 percent failed. In my last shipment a few days ago 2 drives didn't INITALISE and of the 2 drives that Initalised ... one drive in SMART showed YELLOW CAUTION on it, and only 1 drive passed Smart and using HHD had 0 bad sectors at default settings on surface read mode (took about 14 hours).
What is your experience on these drives? Sounds like you have knowledge most of us don't have with these drives.
You said you had two drives that wouldn't initialize. Did you do anything to test them? Ideally, tests like HDDScan and Victoria should be run on uninitialized drives. Once they have been initialized (a partition table created), Windows will restrict some of the abilities of these programs (mainly their ability to directly write to the drive). That said, them failing to initialize is cause for concern.
What program showed a yellow caution warning? Do you know the details of the SMART report? The lack of industry standardization makes SMART values a nightmare to interpret. Many programs (and users) incorrectly interpret SMART information, especially on less common drives. However, these particular drives likely do have good support. If you have the actual SMART report, I'd be happy to take a look and give you my opinion. If you were using an updated version of CrystalDiskInfo, on a drive like this, a caution warning likely is cause for concern.
For the Victoria HD test I left everything at default. That's why I mentioned the 10,000 ms timeout.
I agree with you that even though the timeout is set that high there must be another criterion when this test is run since it picked up ms timeouts less than the 10,000 ms default setting at 9203 and 8532 ms.
Also, you must be right about Windows trying to access the drive causing erroneous slow/weak sectors. That may be the case here with these two high ms timeout results.
Also, I'm running these tests on a USB external HD enclosure, since I have only a couple of laptops, no SATA interface as these tests should be run, and that may affect this Victoria test in taking so long and perhaps providing erroneous results.
The CHKDSK tests that I did on both of these 10Tb drives took about 16 hours and both came back with NO bad sectors/clusters.
Both drives were formatted with the default NTFS 4K allocation unit size or block size before the CHKDSK and the Victoria tests were run.
The 8.33 ms timeout link I mentioned is of course ideal, and it was mentioned purely as relevant information.
I haven't run the HDDScan test yet.
Screenshots of the Victoria test currently running are posted.
You are right about Windows accessing the drive is causing erroneous slow/weak sectors.
The screenshot below with the third slow/weak sector was when the drive was accessed by Windows to see the SMART of the drive.
So, most likely the other two listed weak/slow sectors are erroneous too.
Yes, it is definitely better to use a native SATA port, rather than a SATA-USB bridge. I strongly suspect the reason the test is taking so long is because yours is currently linked at USB 2.0 speeds. No portion of that drive should have sequential speeds anywhere near 27MB/s. It should be in the ballpark of 250-100MB/s.
Those 8-9 second delays are kind of weird. That is a long time. I wonder if the drive may be going to sleep and waking back up. That is a long enough delay that, if it was the drive struggling to read, a drive with TLER (Time-Limited Error Recovery) enabled may timeout and throw an error. I've seen default TLER timeouts of 7-10 seconds. That would also depend on the feature being available and enabled on the drive. Counterintuitively, I've seen it enabled on WD EasyStore drives and available but disabled on some Ultrastars.
Anyway, unless something suspicious shows up in SMART, I'd be inclined to believe those are erroneous delays. They're single blocks, spaced apart. With actual surface damage, it's common to see clumps in LBAs that are close together. If you run any further tests and have to use the USB adapter, try to make sure it's linked at USB 3 speeds. The beginning of that drive should be able to hit 200MB/s, if not more.
You said you had two drives that wouldn't initialize. Did you do anything to test them? Ideally, tests like HDDScan and Victoria should be run on uninitialized drives. Once they have been initialized (a partition table created), Windows will restrict some of the abilities of these programs (mainly their ability to directly write to the drive). That said, them failing to initialize is cause for concern.
What program showed a yellow caution warning? Do you know the details of the SMART report? The lack of industry standardization makes SMART values a nightmare to interpret. Many programs (and users) incorrectly interpret SMART information, especially on less common drives. However, these particular drives likely do have good support. If you have the actual SMART report, I'd be happy to take a look and give you my opinion. If you were using an updated version of CrystalDiskInfo, on a drive like this, a caution warning likely is cause for concern.
You mentioned that you bought ServerPartsDeals and I bought 8 of their 14TB manufacture refurbs, and they passed all the tests, no problems since I bought them. All of them initalized and smart looks good.
I bought 4 of the 12TB of the drives in this thead from GoHardrive over a week ago, and they all initailized fine and smart was good. It was my first dealings with GoHardrive and I was happy. So I bought another 4 12TB after that and it these ones where I had the bad experience that I shared in a few other posts.
I have sent them back to GoHardrive.
If ServerpartsDeals would not initalized or gave smart problems I would have sent them back too.
The Gohardrive THREAD in this room say in big red letters in WIKI or first page of this thread, ...
Community Wiki
Heavily used drives
Sign up for a Slickdeals account to remove this ad.
For the Victoria HD test I left everything at default. That's why I mentioned the 10,000 ms timeout.
I agree with you that even though the timeout is set that high there must be another criterion when this test is run since it picked up ms timeouts less than the 10,000 ms default setting at 9203 and 8532 ms.
Also, you must be right about Windows trying to access the drive causing erroneous slow/weak sectors. That may be the case here with these two high ms timeout results.
Also, I'm running these tests on a USB external HD enclosure, since I have only a couple of laptops, no SATA interface as these tests should be run, and that may affect this Victoria test in taking so long and perhaps providing erroneous results.
The CHKDSK tests that I did on both of these 10Tb drives took about 16 hours and both came back with NO bad sectors/clusters.
Both drives were formatted with the default NTFS 4K allocation unit size or block size before the CHKDSK and the Victoria tests were run.
The 8.33 ms timeout link I mentioned is of course ideal, and it was mentioned purely as relevant information.
I haven't run the HDDScan test yet.
Screenshots of the Victoria test currently running are posted.
You are right about Windows accessing the drive is causing erroneous slow/weak sectors.
The screenshot below with the third slow/weak sector was when the drive was accessed by Windows to see the SMART of the drive.
So, most likely the other two listed weak/slow sectors are erroneous too.