Jump to content
Covecube Inc.
  • 0
robti

What software to test new drive

Question

Hi in the past I just added my drives to the pool, but they were only old drives that were being reused from my htpc, now getting ready to add a 4Tb red and wondered if you all test new drives before adding and what to use for it ?

 

Thanks

 

Robert

Share this post


Link to post
Share on other sites

18 answers to this question

Recommended Posts

  • 0

I believe HD tune has a cost, never used it,  the manufacture often has free tools that you can use, but I usually perform full format which takes several hours, especially on a 4tb HDD,  then I let stablebit scanner perform a full scan, which is several more hours, and let the drive run for a week or so in the server before adding to the pool.  most of my drives have been pulled from external enclosures, as a matter of fact all of my 3.5" drives excluding 1 HGST NAS Drive have been pulled from enclosures,  With that said,  I will typically run the full format and stable scanner before busting open the enclosure, and then repeat the process again when added internally to the server.  

 

Nothing too crazy.  

Share this post


Link to post
Share on other sites
  • 0

I use StableBit Scanner. :):P

 

Once the surface scan gives it a clean bill of health, I add it to the pool.

 

I know a number of people that use SpinRite on disks before putting them in "production." However, SpinRite does not support drives larger than 2TBs.

 

Aside from that, I used to do a full format. That writes zeros to most of the disk, and should flag any issues and fix them to start off with (and may trigger SMART errors).

Share this post


Link to post
Share on other sites
  • 0

..I usually use and run the preclear script from the unRAID forum...not only for used drives but for new buys (three times) as well.

If a new drive fails, it goes the RMA way right away.

See: http://lime-technology.com/wiki/index.php/Configuration_Tutorial#Preclear_Hard_Drives

The script is a simple shell script...I run it when booted from a Fedora Live-ISO.

Share this post


Link to post
Share on other sites
  • 0

Thanks how do I do a surface scan with stablebit?

It should start right away with newly connected drives, or whenever the work window is for StableBit Scanner.

 

 

..I usually use and run the preclear script from the unRAID forum...not only for used drives but for new buys (three times) as well.

If a new drive fails, it goes the RMA way right away.

See: http://lime-technology.com/wiki/index.php/Configuration_Tutorial#Preclear_Hard_Drives

The script is a simple shell script...I run it when booted from a Fedora Live-ISO.

As I suspected, it's reading from the disk, and then doing a full write zero pass to the disks. 

I'm not really a fan of doing this multiple times. Generally, once should be enough. Heck, reading the disks start to finish should be enough to detect any issues with the drive.

Share this post


Link to post
Share on other sites
  • 0

As I suspected, it's reading from the disk, and then doing a full write zero pass to the disks. 

I'm not really a fan of doing this multiple times. Generally, once should be enough. Heck, reading the disks start to finish should be enough to detect any issues with the drive.

I am an ex unRAIDer (have swapped to ZFS), but in the days back then, I found that the guys in the unRAID forum have some very valuable insights on how to treat your disks.

Making sure that the parts are working as expected is vital...that forum is full of evidence, that even brand new disks will fail the test, when run more than once.

I agree that is is time consuming, but there is no harm done to the disk ... it should pass clearly a test with 3 or more runs, when mounted correctly and airflow in place.

From an engineering perspective I think one can agree that a disk that passed multiple runs is potentially less prone to failure, as real-live stress would be lower than that of the test runs.

A new disk that has passed one run but would have failed on later runs will most likely fail earlier than others, while in service.

IMHO it is better to find the issue while you still can RMA the thing...and it is less disappointing (and less risky) compared to when you have a fully employed disk with your valuable data on it.

 

Edit: ..and to what the script does...

yes, it is writing/reading zeros, but that is only one part of the story...it is trying to find if re-allocated sectors/blocks would occur during the process.

If that value increases, this indicates a problem with the disk...in order to "force" this, the writes/reads are necessary....just as in real life use.

Share this post


Link to post
Share on other sites
  • 0

A new disk that has passed one run but would have failed on later runs will most likely fail earlier than others, while in service.

IMHO it is better to find the issue while you still can RMA the thing...and it is less disappointing (and less risky) compared to when you have a fully employed disk with your valuable data on it.

Regardless of the rest, the bit about less risky than finding out issue with data on it is really the key point here, I think. 

Once the data is on the disk, if it isn't healthy, then you risk data loss, and that's never okay.

 

Edit: ..and to what the script does...

yes, it is writing/reading zeros, but that is only one part of the story...it is trying to find if re-allocated sectors/blocks would occur during the process.

If that value increases, this indicates a problem with the disk...in order to "force" this, the writes/reads are necessary....just as in real life use.

Actually, (without having access to the specific script, as i'm not a registered user there), it does a full zero pass of the disk, then does a full sector by sector read (i'm guessing on this part). It then looks at the SMART data for reallocated sectors (which Scanner can see as well).

 

In fact, to duplicate this, you'd write zero's to the entire disk, then let StableBit Scanner do a full surface scan. 

 

And to be honest, doing a full format (which is what I usually do) and letting Scanner "do it's thing" should be essentially the same thing.

 

 

ALso, I know that a few people have asked us to include a full write pass to the disk. It's something that we've talked about and considered. However, it hasn't been a priority to get into StableBit Scanner as it is VERY dangerous if you don't know what you're doing.

But for a power user, this may be a great feature (it will have 2-3 prompts at a minimum, if we do implement it)

Share this post


Link to post
Share on other sites
  • 0

However, SpinRite does not support drives larger than 2TBs.

 

 

Interesting, I did not know that. And from what I've seen, he doesn't make a big effort to announce this on his website or videos.

Share this post


Link to post
Share on other sites
  • 0

Interesting, I did not know that. And from what I've seen, he doesn't make a big effort to announce this on his website or videos.

Yeah, it doesn't appear to be disclosed at all. However, I do know a few people that use SpinRite a lot, and it is a known issue. IIRC, it's an issue with the version of DOS he's using ... but I could be wrong (meaning it could be a limitation in the code he's using).

Share this post


Link to post
Share on other sites
  • 0

Hi in the past I just added my drives to the pool, but they were only old drives that were being reused from my htpc, now getting ready to add a 4Tb red and wondered if you all test new drives before adding and what to use for it ?

 

Thanks

 

Robert

 

If I get a second hand drive, I usually just use DiskPart to do a "Clean All", which writes 0s to the entire disk, wiping any kind of formatting leaving a fresh "uninitialized" disk. Then a quick format to NTFS, and then use a disk check tool to verify surface reads are ok. I use Scanner for that now.

I realize that a "Clean All" doesn't map bad sectors like a full format does, but if you're going to be running a disk surface check afterwards anyway it saves a bit of time. I always found a format of a 4TB drive took longer than a simple clean all (a blind write of zeros to the entire disk).

The other benefit is that the diskpart method clears any hidden partitions you might have missed by simply formatting the existing visible partition.

Share this post


Link to post
Share on other sites
  • 0

If I get a second hand drive, I usually just use DiskPart to do a "Clean All", which writes 0s to the entire disk, wiping any kind of formatting leaving a fresh "uninitialized" disk. Then a quick format to NTFS, and then use a disk check tool to verify surface reads are ok. I use Scanner for that now.

I realize that a "Clean All" doesn't map bad sectors like a full format does, but if you're going to be running a disk surface check afterwards anyway it saves a bit of time. I always found a format of a 4TB drive took longer than a simple clean all (a blind write of zeros to the entire disk).

The other benefit is that the diskpart method clears any hidden partitions you might have missed by simply formatting the existing visible partition.

Well, the "clean all" command should cause the disk to remap/reallocate any damaged/questionable sectors, so you should be good on that account.

 

 

And yeah, it clears all of the "OEM" partitions that may exist on the drive. But just be very careful with this, as you could totally wipe (and make recovery VERY difficult) if you get the wrong disk.

Share this post


Link to post
Share on other sites
  • 0

I have a question relating to this.

 

Not long ago, I had a drive that DrivePool or Scanner, I don't remember which, reported as bad.  I ran to BestBuy, picked up another 2TB drive, replaced it, and went on my way.

 

This weekend, I plugged that drive into an enclosure, and ran ActiveDisk to "kill" it which writes 0's across the entire disk.

 

Given that the process successfully completed writing the 0's, is there any reason to believe the disk is actually good?  Would creating a partition, formatting it, and running a chkdsk suggest whether or not I could trust the disk again?

 

I understand the risks if unduplicated data were to be stored on it, I'm just curious what other's thoughts are.

Share this post


Link to post
Share on other sites
  • 0

Hi

 

I have a few disks that have reported errors over the years I generally take them out full format run vendors utility's on them stress test for a few days then I if they pass I add them back to the system, I think people panic to much when they get a warning in my experience if a drive is going to fail it generally just goes I have had this with spinners and ssd's I have 2 older wd greens with over 450k park cycles and never had a problem with them keep good backups like you should and I don't see a problem.

 

The oldest drive I have is a spinner that is only 2gb I use it in a very old system it must be over 25years old granted it doesn't get everyday use but it's still running.

Share this post


Link to post
Share on other sites
  • 0

I have a question relating to this.

 

Not long ago, I had a drive that DrivePool or Scanner, I don't remember which, reported as bad.  I ran to BestBuy, picked up another 2TB drive, replaced it, and went on my way.

 

This weekend, I plugged that drive into an enclosure, and ran ActiveDisk to "kill" it which writes 0's across the entire disk.

 

Given that the process successfully completed writing the 0's, is there any reason to believe the disk is actually good?  Would creating a partition, formatting it, and running a chkdsk suggest whether or not I could trust the disk again?

 

I understand the risks if unduplicated data were to be stored on it, I'm just curious what other's thoughts are.

 

Depends on the issues, really.

 

In a lot of cases, the full "Zero pass" may fix a lot of issues because it's causing the disk to write to every sector, and triggering the drives' error correction routines.

 

After this, a format isn't really needed (a full format does a zero pass of the partition). And as for a chkdsk, it may not be a bad idea, but again, may not be necessary. But letting Scanner do a full pass of a Surface scan would be telling. As would looking at the SMART data afterwards.

 

And just FYI, I've done this on disks before (specifically that had uncorrectable sector count SMART warnings and damaged sectors). It converted all of the uncorrectabler sectors into reallocated sectors (which is fine). However, after a while, the uncorrectable sector count continued to increase.

The moral is... if issues continue, then RMA or toss the disk. Or use it for non-critical data.

Hi

 

I have a few disks that have reported errors over the years I generally take them out full format run vendors utility's on them stress test for a few days then I if they pass I add them back to the system, I think people panic to much when they get a warning in my experience if a drive is going to fail it generally just goes I have had this with spinners and ssd's I have 2 older wd greens with over 450k park cycles and never had a problem with them keep good backups like you should and I don't see a problem.

 

The oldest drive I have is a spinner that is only 20gb I use it in a very old system it must be over 25years old granted it doesn't get everyday use but it's still running.

 

I've has the opposite experience. Most of the disks I've had go bad on me generally did give SMART warnings prior to failing. 

The batch of bad ST3000DM001's that I had all showed uncorrectable sector counts in SMART. I pulled them immediately.. Two of them completely failed shortly after pulling them (I was using them for testing).

 

However, given how the disks work, there are plenty of ways for them to fail, including suddenly.

 

Though, I've found the most important metric on drives to be their warranty period. 

Not only is that a good indication of longevity, but it usually lines up with the component lifespan info. 

 

 

As for that old 20GB, that's pretty amazing. Though, I'm sure it's slow as dirt compared to the newer drives. That's the trade off, the newer drives pack a lot more tech in them, which means there are more points of failure. :(

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...