Preclear and Failed Drives



Recommended Posts

  • Replies 67
  • Created
  • Last Reply

Top Posters In This Topic

Well folks, Here's that poll I was suggesting. I'm very curious to see the results! And for those of you who aren't running preclear.... What's Wrong With You!!???  :P.

 

I voted for 1 cycle with failed drives.

 

I run 3 cycles, and have had drives fail on the second or third cycle.  I ran a drive with one pass and it was good, I put it into service in my array and then it failed about 2 weeks later.  That is the main reason I run 3 cycles.  I have had a couple of drives preclear on the first cycle OK (not perfect) only to show worse signs on the second and third preclear.

Link to comment

I didn't see the option I wanted, so I edited the poll and added it.  I always run preclear for a minimum of 2 cycles on every drive I use, but I've only ever see a drive fail on the first cycle if it fails at all.  When a drive fails on the first cycle it typically fails on all subsequent cycles as well.

 

I have also seen a few drives that took exceedingly long on the first cycle (like 2x or longer than normal), but still passed, then ran at normal speed on all subsequent cycles.  These drives have all turned out to be fine in normal use.  I can't explain why the first pass took so long.

Link to comment

Suggested options ...

 

- What's preclear

- Ive had a drive fail after 1 or more preclear cycles (please post your experience)

- I've had a drive fail in the first cycle of preclear

- I've had a drive fail in the second cycle of preclear

- I've had a drive fail in the third or higher cycle of preclear

- I routinely run 1 preclear cycle and never had a drive fail during preclear

- I routinely run 2 preclear cycles and never had a drive fail during preclear

- I routinely run 3 or more preclear cycles and never had a drive fail during preclear

 

 

Link to comment

Well folks, Here's that poll I was suggesting. I'm very curious to see the results! And for those of you who aren't running preclear.... What's Wrong With You!!???  :P.

 

I voted for 1 cycle with failed drives.

 

I run 3 cycles, and have had drives fail on the second or third cycle.  I ran a drive with one pass and it was good, I put it into service in my array and then it failed about 2 weeks later.  That is the main reason I run 3 cycles.  I have had a couple of drives preclear on the first cycle OK (not perfect) only to show worse signs on the second and third preclear.

 

I'm dealing with my first failed drive and this is what I've seen. 1 successful preclear, then during the rebuild I had all sorts of failures. Think I'm gonna run a few preclears on another spare drive I have.. phew good thing I keep a few around.

Link to comment

I have also seen a few drives that took exceedingly long on the first cycle (like 2x or longer than normal), but still passed, then ran at normal speed on all subsequent cycles.  These drives have all turned out to be fine in normal use.  I can't explain why the first pass took so long.

 

I've had one drive that was slow on the first preclear pass (it took about twice as long as normal) and was just as slow on the second pass, so I took it back to the dealer and got a replacement.

 

Stephen

Link to comment
  • 2 weeks later...

In my case a brand new Seagate 2TB ST32000542AS drive passed the first preclear, but with some sectors re-allocated.

The second preclear it re-allocated LOTS of sectors, started making LOTS of funny noises, and failed completely.

 

 

 

 

Lesson learned. On my new drive I just got I'm going to do 3 clears instead of the single clears I did on my older drives that have been in service for months.

Link to comment

This poll reminds me of my Hitachi 7k2000 parity drive that passed two pre-clears, was in service 6 months, threw 5-10 errors, smart reported showed reallocated sectors, swapped in a spare drive, pre-cleared the erroring drive about 10-11 times.  It failed all but the last two, but did pass the final two pre-clear attempts.  I put it back into service as a data drive and it has been working perfectly for several months.

 

So I guess I don't see my choice.  I preclear until the drive can pass two consecutive pre-clears, or until it reaches a failure criteria that would let me RMA it.  Even though my Hitachi was failing one pre-clear after another, it would still pass the stupid Hitachi drive analysis test, so it wasn't eligible for RMA.  So I was going to preclear that drive until it fried if that is what it took.

Link to comment

I'll take a long data re-build a good clear for me. If the drive is going to fail right away, it will during a re-build. A drive can fail at any given time.

 

I don't think that is true.  A data rebuild will write to the entire drive, so that's good.  But it won't read from it.  A data rebuild followed by a parity check would be better in my opinion, but still not as good as a full preclear (which is a read, then a write, then a read again).  The preclear also tests both sequential and random writes/reads, whereas the data rebuild + parity check would only test sequential writes/reads.

Link to comment

Great post/poll. I will not vote just yet...

 

I started my unRAID & preclear_disk experience as followed. I precleared many old drives which I will use at the end of my 24 drive unit. Once I exceed all available slot with 2TB drives I will then swap out my older drives. Squeezing my money worth for what I paid for these puppy's back in the day. (P.S. all Hitachi 7200k drives. 160Gb, 250Gb, 320Gb)

 

I was preclearing these old drives with 3 Passes, only one drive showed reallocated sectors from a very lond time ago and no new additions. So I thought this was the recommended. I then started purchasing my new Hitachi 2TB, first was a 7200k for parity and precleared it with three passes all was well, the second Hitachi was a 2Tb green drive, also 3 passes. All these drives were done in an old foxconn MB. When I moved them to my new supermicro MB with SAS controllers, i decided to run one more pass on each, I was not aware of the -D option in preclear and ended up with now SMART data in the reports... So I ended up running another pass on each with the -D option to get the SMART data in the reports. ALL still came up fine. Then I read a post here in the forums about what does 1 preclear pass equate to for a harddrive lifespan (workload)... i think most of us were surprised by Joe L.'s anwser. So my next (2)2TB Hitachi green drives I only ran 2 passes on and all is still well. I am waiting on 4 more 2TB Hitachi's and plain to do the same 2 FULL passes on each, if I see anything out side of 0,0,0,0's then I will apply more passes. And then I will vote in this poll.

 

Thank you for the great script Joe L., great post/poll guys! and love the forum as well as the maturing product unRAID! rocks.

Link to comment
  • 2 months later...

Somehow I never noticed this thread until now.

 

I run 3 or more preclear cycles on my drives and I've seen drives pass the first cycle but failed hard in the second cycle. The third cycle confirmed that the drives were toast. It even failed the smart short test with read failures!

Link to comment

Somehow I never noticed this thread until now.

 

I run 3 or more preclear cycles on my drives and I've seen drives pass the first cycle but failed hard in the second cycle. The third cycle confirmed that the drives were toast. It even failed the smart short test with read failures!

 

Shhhhh, it a secret poll....  ;D

Link to comment

 

Somehow I never noticed this thread until now.

 

I run 3 or more preclear cycles on my drives and I've seen drives pass the first cycle but failed hard in the second cycle. The third cycle confirmed that the drives were toast. It even failed the smart short test with read failures!

 

Shhhhh, it a secret poll....  ;D

 

I had seen the poll, but somehow never got to post here. The secret's out!

 

I got used to runnning two preclears on drives 'new' to my tower, and three on brand new drives. Have seen a few drive failures at the second cycle. Now just run the two cycles, and decide. Something like 16-18 drives total precleared since my March build up.

 

Link to comment
  • 4 weeks later...

On my 1st build (4220 started at 11 drives, and maxed out with 18 data drives a mix of 1, 1.5, and 2 TB drives, running unRAID 4.5.6) I did not run pre-clear, and have had one drive fail in 16+ months of operation. (Running 24/7 98% of the time.) But you all have me thinking that on this new build (4224 starting with 12 data + plus parity all are 3TB drives, running, unRAID 5b12) I should run pre-clear.

I need to read more about this.

 

Ok, in this thread http://lime-technology.com/forum/index.php?topic=2817 It states to use 1.6 or later for 5b4, but when I down loaded preclear, it has no ver. designation on it, how do I know I am getting the correct version?

Link to comment

The preclear program itself will tell you, execute the -v switch and it will tell you the version:

 

./preclear_disk.sh -v
./preclear_disk.sh version: 1.9

 

(don't tell Joe L. I haven't updated in a while!)  ::)

 

If you have any other questions, the -? switch should answer them.

 

./preclear_disk.sh -?

 

Joe L. provided a LOT of documentation for this, and it just keeps getting better!

 

Bruce

Link to comment
  • 3 weeks later...

As a way noob all I can say is holy crap, 3 preclear cycles!!!  :o  I just finished my first single cycle preclear on a 1TB Seagate drive and it took 25 hours or so.  I would seem that 3 cycles might be worth the effort, so I guess I'll have to keep running my WHSv1 setup while I get things started on my new unRAID setup.  Good thing I have a spare old PC or two down in the basement to play with.

Link to comment
  • 2 weeks later...

As a way noob all I can say is holy crap, 3 preclear cycles!!!  :o  I just finished my first single cycle preclear on a 1TB Seagate drive and it took 25 hours or so.  I would seem that 3 cycles might be worth the effort, so I guess I'll have to keep running my WHSv1 setup while I get things started on my new unRAID setup.  Good thing I have a spare old PC or two down in the basement to play with.

 

3 cycles is my preclear number, and on the 3TB Hitachis it takes about 105 - 110 hours.  Found a few duds this way, so I can wait a week for my drives to be "cleared" for duty.

Link to comment
  • 1 month later...

I didn't see the option I wanted, so I edited the poll and added it.  I always run preclear for a minimum of 2 cycles on every drive I use, but I've only ever see a drive fail on the first cycle if it fails at all. 

 

That may be because you haven't tried 6, 7 or even 10 cycles that you haven't seen one fail after two passes ;)

 

The last drive I sent back failed on pass six.  Obviously it could get out of hand and you could preclear for a month, but I figure ten solid days of preclear in not-so-great temperature conditions (drives reach 50C or so) is a pretty good test as to whether or not it will be reliable for a few years.

 

I'd rather endure the ten-day wait than have more than one drive fail at the same time in my array.

Link to comment

I didn't see the option I wanted, so I edited the poll and added it.  I always run preclear for a minimum of 2 cycles on every drive I use, but I've only ever see a drive fail on the first cycle if it fails at all. 

 

That may be because you haven't tried 6, 7 or even 10 cycles that you haven't seen one fail after two passes ;)

 

The last drive I sent back failed on pass six.  Obviously it could get out of hand and you could preclear for a month, but I figure ten solid days of preclear in not-so-great temperature conditions (drives reach 50C or so) is a pretty good test as to whether or not it will be reliable for a few years.

 

I'd rather endure the ten-day wait than have more than one drive fail at the same time in my array.

 

10 cycles?!  You crazy! :P  Running a drive through 10 cycles of preclear in a 50C environment is more likely to kill the drive than test it.  I prefer to test drives in more realistic conditions.  unRAID is generally used as a mass data store/archive.  Drives aren't accessed constantly, like they would be in a web server or large corporate environment.  Temperatures should be much lower as well (45C is my personal cutoff - if I see any drive above 50C I would immediately abort the test and focus on improving cooling).  Hence, 2 - 3 passes of preclear in a temperature range of 30C - 45C is just about right in my opinion.  A drive that fails after 6, 7, or 10 preclear cycles could have very well lasted months or years under normal unRAID conditions.

 

If you are that worried about simultaneous drive failures, then I suggest you keep both a warm and cold spare handy and have a full off-site backup of your critical data.  In my opinion that is much better insurance against data loss than beating drives to death at an early age.  Remember that drives replaced under warranty are often refurbished drives, which can end up being even less reliable than the drive that failed after 6 preclears.  It is best to get the most life out of your new drives and then capitalize on the warranty after a year or two of faithful service.

Link to comment
  • 2 weeks later...

 

10 cycles?!  You crazy! :P  Running a drive through 10 cycles of preclear in a 50C environment is more likely to kill the drive than test it.  I prefer to test drives in more realistic conditions.  unRAID is generally used as a mass data store/archive.  Drives aren't accessed

 

I agree it is wacky to do 10 passes, that's why I do 7 passes or ten days, whichever takes longer ;)

 

The 50C plus environment is not by choice - it happens in external drive enclosures where I cannot control cooling well enough.  If I want to RMA one of those drives before it causes data loss, I basically need it to fail while it's still new (or new to me).  As soon as I take it out of the case I can no longer RMA it... and wouldn't you know, where I am it is often the case than an external USB drive is cheaper than an internal one?  Not always, but there ya go... the world is full of strange occurrences.

 

Anyway my theory has recently proven me wrong.  I had a Hitachi 7200rpm drive in an external case that passed gruelling 11 days and 7 passes with spikes over 60C during preclears, and it passed with flying colours - only to fail as soon as I tried to rebuild the array with it in place!

 

Anyway, I think I'll still stick with 7 as it has helped weed out four bad drives in the past few months, a couple on passes 1 or two, and one one pass 5, another on pass 6.  In those four cases the preclear regimen was enough to send each drive into RMA land with SMART errors and bad sectors galore.

Link to comment
  • JorgeB unpinned this topic

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  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.