Is there a point at which slow SpinRite performance means the drive is bad?

[brado2049]

I am not new to SpinRite, but new to successful SpinRite runs — long story, hardware config challenges. I am past those challenges, and now running Level 3s on a bunch of 3TB and 4TB hard drives of mine. My first run was on a Toshiba 4TB SATA drive — it took about 12-13 hours to complete had consistent progress throughout. But now I’m running a Level 3 on a 3TB Seagate Barracuda SATA drive, and the thing has been running over 15 hours and only 1.2% complete. It is estimated over 1195 hours (49+ days) remaining. Questions:

1. Is this kind of performance normal for SpinRite, a month-and-a-half for a 3TB drive?
2. Is there a point at which you should kill the current run of SpinRite, and try running again (perhaps with different options)?
3. How do you know a hard drive is just bad and needs to be thrown away?

In general, I’m trying to understand when I’m within the realm of normal SpinRite behavior and should let things continue to run, or when I’m seeing something abnormal and should do….???

Thanks so much for your help!

 

[ColbyBouma]

That drive is definitely struggling. If you scroll through the screens (left and right arrow), it'll give you more information such as number of retries.

 

[brado2049]

@ColbyBouma - I’m not sure exactly what I’m looking for, but here are screenshots of all the screens.

 

[ColbyBouma]

Thank you. If you have any important data on that drive, stop SpinRite and try to make a copy of the data.
Here are some things I noticed:

1. The Real-Time Activities page has 8 counters. On healthy drives, all of those are usually 0. This drive has 4 counters above 0, the most worrying of them being "not recoverable". That means SpinRite tried to recover data from a sector, but was unable to get 100% of the data, so it gave up and overwrote that sector with whatever it was able to find.
2. The yellow "Waiting for drive" in the top-left corner of several screenshots means SpinRite sent a command to the drive, but it hasn't responded yet. Healthy drives never do this.
3. The empty S.M.A.R.T. System Monitor page is unusual.
4. The ST3000DM001 is known to have exceptionally high failure rates. https://en.wikipedia.org/wiki/ST3000DM001

 

[peterblaise]

Right on @ColbyBouma!

- - - - -

After backup, and especially if backup fails to get everything:

SPINRITE NORAMTEST LEVEL 5 DYNASTAT 1 NOREWRITE

... just to see if the drive itself is recoverable.

It's probably a test drive from here on out.

- - - - -

Useful command line options:

DYNASTAT 0 recovers nothing, good for blankable drives to 'quickly'
plow through with a surface integrity refresh.

DYNASTAT 1 prevents SpinRite from wasting 4 extra minutes
recovering nothing additional.

NOREWRITE prevents SpinRite from writing zeros in unrecoverable
areas so you can revisit and try again by any other means.

NORAMTEST bypasses a quirk where drive enumeration produces
ghosts, not your problem here, but who knows?

SKIPVERIFY bypasses pre-test before actual Level testing, not your
problem here, but good to know just in case the drive stops being
recognized by SpinRite at all.

- - - - -

Thanks for following up, and let us know how that drive behaves next.

 

[brado2049]

@ColbyBouma @peterblaise — thanks so much for your responses! Additional status — I kept it running overnight, and a day later, not much progress. I’m attaching another set of screenshots from this morning showing pretty much the same state of affairs. A note on this drive (and all of my drives) — I have no data I want to recover at all on any of them. A repair and refresh is what I need. @peterblaise ’s command line appears to be a good next step, so I’ll kick that off. But I noticed this was a Level 5. It got me to wondering — if I want to recover no data at all, but repair and refresh a drive, what are the options I should be using?

One last note — I’m using a ZimaBoard with the SATA-Y cable and two drives connected, and I set SpinRite running on two SATA drives and the first 4TB one finished fine. This drive I’m having problems with is the second one. I checked the Zima web site and the doc on the SATA-Y cable and it is supposed to be compatible with the ZimaBoard, but I wondered if may I ought to revert back to the single SATA cable which came with the ZimaBoard just in case there was possibly some power fluctuation being introduced b/c the Y cable was powering two drives instead of one. Has anyone had any problems or behavioral differences when using the SATA-Y cable and two drives vs just a single drive?

Ok…I’m of to restart with @peterblaise ’s options. More status as it comes…thx so much for your help!

 

[PHolder]

You're potentially investing more personal time in a drive that is not worth the investment. If you envision your time is worth as little as $10/hr you could reinvest that money in buying a new reliable drive. Unless this drive has emotional or sentimental value to you, it's time for it to permanently retire.

 

[Tazz]

if I want to recover no data at all, but repair and refresh a drive, what are the options I should be using?

Personally I would run a

spinrite /dynastat 0

Level 4. Level 5 if time does not matter.
If time does matter then a Level 3 would give a good idea of how bad it is.

Also, if it's testing bad at the beginning of the disk, start SpinRite at the 50% mark and then go back to get the first part later.

 

[brado2049]

@PHolder — you make a good point — the value of time. That’s actually what’s behind these questions, which aren’t so much about this one apparently faulty drive (that’s just made for a context to ask them), but rather understanding SpinRite and its behavior and output. Mastered once, that’ll carry me indefinitely.

The bigger picture is this — I rarely (pretty much never) have data recovery needs due to drive failure. I run direct-attached RAID arrays which do continuous backups to a NAS RAID array. The only time I’ve ever lost data was inadvertently when migrating to a new drive array (my stupidity). But data loss from drive failure is a very low risk for me — I’d need to lose multiple drives on multiple arrays simultaneously for that to happen. Never had it happen in over three decades.

But what I do have are hard drives that fail occasionally. I have a stack of 13 SATA 3TB and 4TB hard drives (none are specifically for NAS) that I’ve pulled over the years when the array software indicated these drives were failing. So I’m running them through SpinRite now.

@Tazz — hey, great minds! After taking guidance from @peterblaise ’s post, I decided that I needed a point of reference with this stubborn 3TB Seagate drive — I have a stack of others, same drive model, so I gunned up another one doing a Level 3 Dynastat 0 and it immediately started speeding through faster than my Toshiba 4TB — estimated an 8 hour run (I’ve attached a screenshot).

So this takes me back to one of my original questions — how do you know when a drive is no good anymore? What is the specific criteria (or is there)? I have a stack of these to plow through now, a little wiser than when I started (thanks to the folks here on the forums — thanks everyone!) Hopefully, most of these will be just start / set watch alarm / return and start the next. But it would help to know exactly when to consider a drive dead.

Thanks!

 

[PHolder]

how do you know when a drive is no good anymore?

That's a statistics problem hiding in sheep's clothing In theory you could get a brand new drive plug it in, and your system could get fried by a lightning strike 5 seconds after you wrote some important file to the drive for the first time. Statistically that probably almost never happens, but my point... if I even have one... is that a drive is worth exactly as much trust as you have in it retaining your important data for a lengthy enough time to meet your expectations. I gather that's not very helpful to you, but I think, it's all I got. If a drive is acting wonky, or unreliable, you're setting yourself up for hurt if you put too much trust in it.

 

[brado2049]

@PHolder - thanks for the reply. Forgive me for not being more clear. The context of the question is relative to SpinRite use. Question restated: Under what conditions is it worthwhile to continue trying alternative options and/or continuing to run SpinRite on a drive vs. just canceling / exiting out of SpinRite and tossing a drive in the trash? I imagine there must be some statistic, count of errors, timeouts, length of time running, ...something... which makes it a reasonable conclusion that a drive is usable, and/or a reasonable conclusion that a drive is not usable. If there is no threshold or standard, and it is all governed by The Force, then I am asking any SpinRite Jedis out there for your own personal sense of it when you run SpinRite: what criteria do you use to determine a drive can be saved and continued to be used vs. it is dead?

 

[DanR]

If there is no threshold or standard, and it is all governed by The Force, then I am asking any SpinRite Jedis out there for your own personal sense of it when you run SpinRite: what criteria do you use to determine a drive can be saved and continued to be used vs. it is dead?

Unfortunately, there is no good/bad black/white yes/no answer. It comes down to a combination of judgement, common sense, and experience.

If SpinRite gets stuck and unable to move on -- that is bad.

If there any U's or B's in the Graphics Display Screen - that is bad. If there are only some R's that could be OK as it indicates SR recovered the data.

If there are errors at the bottom of the Real Time Activities screen - that is bad! Good drives will have zero errors; bad drives lots of errors. And then there is the gray area in between . . .

If the Detailed Technical Log screen has a very small scroll bar over on the RH side, that is bad! It means there is a LOT of data (likely error data) in the DTL to scroll thru - bad.

If the S.M.A.R.T. data sccreen is blank - that is bad.

The drive you initially posted about is bad. It is beyond SpinRite's ability to "fix" it. SpinRite can do things that seem like magic. But SR does have its limits.

In regard to the drive in your OP, there are a couple of things you might try for info and experience - if you wish to spend the time.

1) Try restarting at 2%, or 5%, etc to get past the initial trouble spot and see what happens

2) Try a DynaStat 0 run. Command line: "spinrite level 2 dynastat 0" and see what happens.

Bottom line: The drive in your OP is toast. It cannot be fixed by SpinRite. It is not to be used ever again. But as a learning tool for playing around . . .

 

[brado2049]

@DanR -- great post, thanks! That gives me a good guideline. A few of those were the mental notes I had gleaned from comments throughout the thread. I indeed had concluded the same about the original drive in question, especially after being able to compare it to the SpinRite performance and output on other drives of the same model.

Dare I ask the logical follow-up -- to what lengths is everyone going to destroy bad hard drives, just the trash can, sledgehammer, drilling holes, or thermite? LOL....can't wait to hear...Thanks!

 

[Tazz]

how do you know when a drive is no good anymore?

Adding to what the others have said, if the SMART info is accessible - Reallocated Sectors and Pending Reallocated Sectors are usually the beginning of something that doesn't get better. The rest of the SMART stuff is vendor specific on whether or not the numbers mean what it looks like they mean - large numbers don't necessarily mean bad things.

Dare I ask the logical follow-up -- to what lengths is everyone going to destroy bad hard drives, just the trash can, sledgehammer, drilling holes, or thermite? LOL....can't wait to hear...Thanks!

I like to take them apart and pull the magnets out but I'm not allowed to do that anymore because "You have too many that's taking up too much space and they're all stuck in a jumbled ball." - my wife.

Break the PCB and cut the ribbon cable then pound a nail down through the top should be good enough unless you think someone is trying to gather intelligence on you.

Just do something so that the average person who may find it and plug it in can't get access to it. Again, unless you have a stalker.

 

[brado2049]

@Tazz -- brilliant, thanks!

 

[brado2049]

Ok @DanR (or anyone), I have another good example to get your read on things. I am including another set of screenshots on a different drive, same model (Seagate Barracuda 3TB) as the first one, and here’s what I see:

• On the Graphic Display Screen, shows SpinRite has been running 3.5 hours, estimates 95 more hours.
• No errors shown on the Real-Time Activities screen.
• On the S.M.A.R.T. System Monitor screen, there appear to be some errors (uncorrectable).

So, very slow SpinRite processing, and it appears some uncorrectable errors (if I’m reading that right). @DanR (or anyone), what’s your take on this drive? If the estimates are right, 95 hours is ~4 days — if this is your hard drive, what do you conclude and do? Is this drive likely salvageable and worth completing, or does this drive belong alongside the lost Atari 2600 E.T. cartridges in a New Mexico landfill? What say you? (yes, channeling Aragorn from the LOTR movies… :-D )

 

[PHolder]

Seagate Barracuda 3TB

From the model # on your screen cap: https://en.wikipedia.org/wiki/ST3000DM001

Backblaze, a remote backup service company, observed that its ST3000DM001 drives have failed at rates far higher than the average of other hard drives. Only 251 of the 4,190 ST3000DM001 hard drives placed in service in 2012 were still in service as of 31 March 2015.

 

[brado2049]

@PHolder — yep, @ColbyBouma mentioned that earlier in this thread. I’ve still got some working ones though, two of the four I’ve run SpinRite on so far have completed fine. One we know is bad, the other is still running and the subject of my most recent set of screenshots above. The general question on that drive still holds — does the data indicate a drive that isn’t worth pursuing, or one that still may be corrected?

 

[peterblaise]

Sadly, Backblaze has no published standards by which they replace a
drive or call a drive 'failed', that is, if they replace a drive because of
S.M.A.R.T. or other reports, WITHOUT FAILURE, or if the drive
actually becomes unable to read ( and write ) data, and then and only
then it is replaced <-- not likely.

I presume Backblase replaces drives BEFORE actual failure, and may
never retest after that, and as such, they may never know if the drive
would have eventually failed, or not.

As such, Backblaze's recommendations are most appropriate for large,
multiple-installation, financially ROI-based return on investment
organizations looking for assurance that they are avoiding LIKELY
risks, but they may not be reporting individual REALIZED risks.

I'd love to test Backblaze's deprecated drives to see if some are
perfectly fine by my standards, or if they really, really, really are a
waste of time and energy because critical failure is imminent or has
already happened.

Who knows?

- - - - -

So for those of us out here working with onsies and twosies, not a
server room of thousands of drives, we gotta deal with maybe the
x% of the drives that prove reliable, versus the y% that have failed.

There's no way to correlate our drive to Backblaze's report.

So, what do we do?

I ardently and continuously WATCH the

SpinRite 6.1 [ Real-Time Activities ]​

... screen to see if any region or sector is slow to read or write:

Illustrated ( perhaps subtly? ) here:

​

And, of course, watch the events enumerated on that screen:

command timeout: 0 command aborted: 0 ​

comm/cable errs: 0 not recoverable: 0 ​

minor troubles: 0 sect neverfound: 0 ​

dynastat recovr: 0 defective sectr: 0 ​

... and toggling over to the [ Detailed Technical Log ] screen, for
example:

And through those views, circumnavigate the drive's read and write
performance.

Plus, listen to the drive, watch it's temperature, and so on, to become
intimately familiar with 'normal' and acceptable behavior, so I know
what unacceptable behavior looks like, sounds like, and feels like.

And compare to other drives, other makers' drives, and other models.

Eventually, we become familiar with 'good' versus 'bad' regardless of
the success of write-read tests themselves - I have bad drives that
have no failures, they pass all tests, but some of these supposedly
'good' drives are so sloggy that I can't tolerate waiting for them
anymore.

SpinRite is not a pass / fail test for whether or not we will be happy
with a drive.

That's our responsibility to develop the skill to know the difference.

So test those drives, end-to-end, and mark 'em [ TEST ONLY ] for
crapppy ones, and press the winners into service.

Let us know what you do.

Thanks.

 

[DanR]

@DanR (or anyone), what’s your take on this drive? If the estimates are right, 95 hours is ~4 days — if this is your hard drive, what do you conclude and do? Is this drive likely salvageable and worth completing, or does this drive belong alongside the lost Atari 2600 E.T. cartridges in a New Mexico landfill? What say you?

I would say the drive may be bad.
- SpinRite is clearly stuck
- The DTL screen shows a very small scroll bar on the right, indicating lots of data (errors?) in the log - NOT good
- The drive is a Seagate; hence some SMART errors are not unusual for a good drive
- However, the RTA screen shows NO errors at the bottom - Good!!!

I note the the RTA screen shows a sector size of 32768 which is the maximum, and quite possibly too aggressive for this no longer optimum and likely now sensitive drive. You might consider starting SpinRite with the XFER command line token with progressively smaller values for progressively gentler touches, to see if something might work. Another learning experience!

Examples:
spinrite xfer 16384
spinrite xfer 8192
spinrite xfer 4096
Etc.

The suggestions in my previous post, for skipping the bad spot or trying DynaStat 0, for learning and experience would apply here to this drive.
But try the XFER options first!