Jump to content
DoeBoye

Preclear and Failed Drives


68 posts in this topic Last Reply

Recommended Posts

I think I'll be switching to the badblocks + pre-clear method from here on out though.

I run monthly badblocks read tests on each drive once a month.

The day of the month is the disk I test.

i.e.

disk1 - 1st of month.

disk2 - 2nd of month.

disk3 - 3rd of month.

 

etc, etc,

up until the 27th.

I run monthly parity checks' on the 27th.

I do it on the 27th in case it rolls to the next day.. so that in February, the next day after 28th is the first and I'm testing disk1.

I.E. this leaves almost 48 hours for a parity check to finish.

 

What I found is that I have to spawn a keep busy process on the md device. touching a file every minute while I do a badblocks read test on the /dev/disk/by-id/ device so that it does not spin down.

I kill the background process at the end of the job and all is good.

 

This gives me monthly smart and badblock reports along with a monthly parity test.  Hopefully this will weed out any bad or aging drives before they do fail.

Share this post


Link to post

Wow, you have WAY more hard drives than I do, and are WAY more thorough than I am.

 

Now I feel like a jerk for not running monthly badblocks tests on the drives.

 

Thanks for making me feel like a jerk!

 

Now I know what I'll be doing this weekend.

Share this post


Link to post

You shouldn't feel like anything! LOL! Seriously. It all depends on how much you value your data and your recovery time.

I also do a monthly filelist generation so I know what's on a disk should I loose it and I plan to do monthly md5sums on all new files so I can  test the files on a disk should I have to recover it.

This all stems from the one time I did an unRAID upgrade. All of a sudden it decided to 'upgrade my parity' and during the parity upgrade, one of the disks had bad blocks so the upgrade failed. Which meant my disk could no longer be recovered.  (FWIW, I was always running monthly parity checks.. but this one time... I got bit!)

I had to use DDRESCUE to copy the disk forward,  skipping all bad sectors,  then copy the disk backward retrying all failed sectors.

I recovered the whole disk except for 1 bad sector. After that I didn't know which file might have been bad. I suppose I'll hit it in a movie file one day and I'll re-rip that movie, in the meantime, forewarned is forearmed!

Share this post


Link to post

Just picked up 5 new drives and had one so far fail on the first preclear. Still waiting to see how the others turn out. I see value in running extra preclear cycles but I do think there is a point where it just becomes excessive and unnecessary. This is strictly my opinion and if you run many cycles than good for you just not my cup of tea.

Share this post


Link to post

Will a desktop drive thats precleared perform as a long as a precleared NAS grade drive?

The question you need to ask is "what exactly is the difference between a "desktop" drive and a "NAS" drive?

 

My theory... (and just a theory)

  the drive sold for RAID use cost more and often has a longer warranty. 

  (the additional cost covers the expense involved in replacement during the longer warranty period)

  The drive sold for RAID use "may" have undergone a more through final test on the production line.

  The drive sold for RAID use "may" have different firmware to not spin down as often.  Or not try to re-read a sector as many times when they encounter an un-readable sector.

 

  The motor, drive bearings, drive heads, and electronics are nearly identical.  (Why complicate stock issue when there is no real need to have a high tolerance part, and a lower tolerance part.)

 

Once you get past the early part of the "bathtub curve" I expect both types of drive to last as long in an unRAID server.

Share this post


Link to post

So...I just had a 2TB disk fail a pre-clear (one cycle, took 100 hours).  Do I try to preclear again, RMA it (and tell them what?), or throw it in the trash and buy a new HDD?

 

Share this post


Link to post

So...I just had a 2TB disk fail a pre-clear (one cycle, took 100 hours).  Do I try to preclear again, RMA it (and tell them what?), or throw it in the trash and buy a new HDD?

 

RMA it, the manufacturer might have you run a test and send them the results. If the drive passes, continue pre clear until the drive fails, then RMA it.

 

You can post the pre clear results, so how it failed can be determined.

Share this post


Link to post

I have a few disks I am looking to move from my processing server over to my unraid server. I have never had any problems with them thus far and wanted to hear some opinions on running a single preclear instead of multiples. I usually run 2 or 3 preclears on new drives but I was thinking it may be OK to just run a single pass on the drives and assuming nothing jumps out I would be fine. Any thoughts?

Share this post


Link to post

I have a few disks I am looking to move from my processing server over to my unraid server. I have never had any problems with them thus far and wanted to hear some opinions on running a single preclear instead of multiples. I usually run 2 or 3 preclears on new drives but I was thinking it may be OK to just run a single pass on the drives and assuming nothing jumps out I would be fine. Any thoughts?

The reason for several preclear cycles is to get past the early failure part of the "bathtub curve"

http://www.weibull.com/hotwire/issue21/hottopics21.htm

If the drive has been in operation for a while, then 1 peclear cycle will be fine.

Share this post


Link to post

Awesome thanks for the feedback Joe. That was my assumption as well but I am glad to hear it from someone more knowledgeable on the topic than myself.

Share this post


Link to post

So...I just had a 2TB disk fail a pre-clear (one cycle, took 100 hours).  Do I try to preclear again, RMA it (and tell them what?), or throw it in the trash and buy a new HDD?

 

RMA it, the manufacturer might have you run a test and send them the results. If the drive passes, continue pre clear until the drive fails, then RMA it.

 

You can post the pre clear results, so how it failed can be determined.

 

Thanks. I'll save the logs.  I went ahead and started a second pre-clear, just to see and it is *still* running (after 150 hours), most of that time spent in the post-clear read in (which probably just means a bunch more errors).  I'll have to see if I can find the receipt for this one.

Share this post


Link to post

So...I just had a 2TB disk fail a pre-clear (one cycle, took 100 hours).  Do I try to preclear again, RMA it (and tell them what?), or throw it in the trash and buy a new HDD?

 

RMA it, the manufacturer might have you run a test and send them the results. If the drive passes, continue pre clear until the drive fails, then RMA it.

 

You can post the pre clear results, so how it failed can be determined.

 

Thanks. I'll save the logs.  I went ahead and started a second pre-clear, just to see and it is *still* running (after 150 hours), most of that time spent in the post-clear read in (which probably just means a bunch more errors).  I'll have to see if I can find the receipt for this one.

Most manufacturers have a web site where you can enter the serial number of the drive and determine if in warranty.  You probably do not need a receipt.

Share this post


Link to post

Badblocks vs preclear, which is the better stress testing tool and why?

Share this post


Link to post

Badblocks vs preclear, which is the better stress testing tool and why?

Use both, see here:

http://lime-technology.com/forum/index.php?topic=23792.0

 

Personally, I do NOT trust badblocks to identify bad blocks.  In fact, from my own experience in that thread, it is nearly useless, but it does exercise the disk.

If you do use only badblocks, you MUST also look at pre and post smartctl reports for sectors pending re-allocation and sectors re-allocated..

 

Badblocks does not preclear a disk.  (The disk will still be considered as non-zeroed by unRAID and unRAID will zero  the disk once more if you are adding that disk to an existing parity protected array.  During this time, the array will be off-line and no shares will be accessible.)

 

Joe L.

Share this post


Link to post

Running 6x 2 TB WD Greens, a 3TB WD Red for parity and a 1TB WD Green for cache.

 

Any drive I add to it is pre-cleared once because it's compulsory, otherwise I probably wouldn't bother. I'm not going to spend days stress testing a drive to within an inch of it's life because that doesnt represent the kind of load I put on them and besides if a drive fails i'll replace and rebuild.

Share this post


Link to post

I have 3 x 2TB WD20EARS drives that I had in my Windows Home Server.  The server died and I'm moving to unRAID.  I remember running some kind of alignment utility on these drives a long time ago.  Do I need to do anything beyond the default command to preclear them?  I also have 4TB and 2TB WD red drives that will be part of this install, but they currently have all my data on them.

Share this post


Link to post

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.