(Solved) 6 day Preclear, only 1/3 done?

[vjmcdonnell]

I am preclearing a Seagate 3tb HD to use as my parity and its been running for 6 days so far.

 

I know I choose three cycles just to be sure and am worried it will be another 12 days.

see attached screen shot.

 

[garycase]

Wow.  How is this drive connected?  (a PCI card by any chance?)

 

I'd be inclined to let the first cycle finish -- then copy the preclear reports from the flash drive and see if there are issues (seems likely, unless you have a very slow interface to the drive) ... and probably just abort the other 2 cycles.

 

[bkastner]

That can't be a good sign.

 

Here is a report from the last 3TB drive I cleared:

 

== invoked as: ./preclear_disk.sh -A /dev/sdg

==  WDC WD30EZRX-00DC0B0    WD-WMC1T0093370

== Disk /dev/sdg has been successfully precleared

== with a starting sector of 1

== Ran 1 cycle

==

== Using :Read block size = 1000448 Bytes

== Last Cycle's Pre Read Time  : 11:47:58 (70 MB/s)

== Last Cycle's Zeroing time  : 7:01:40 (118 MB/s)

== Last Cycle's Post Read Time : 12:47:28 (65 MB/s)

== Last Cycle's Total Time    : 31:38:05

==

== Total Elapsed Time 31:38:05

 

This was on a PCIE SATA card. As Gary mentioned, if it's a PCI card it would be slower, but 18 days seems a tad excessive.

 

Have you ever cleared a drive on this system before? Was it a brand new drive, or a recycled drive from somewhere else? Again, as Gary mentioned, I can't imagine this drive is in great shape (at a guess).

[vjmcdonnell]

I'm a idiot.

Was trying to preclear in a drive in the array.

Changed to correct drive and flying along.

 

[trurl]

I thought it wouldn't let you preclear an array drive. So you wiped one of your data disks then?

[vjmcdonnell]

No data is still there. Lucky me.

[garycase]

No data is still there. Lucky me.

 

Don't be so sure about that.  Have you actually LOOKED at the contents of the drive?  Go to \\Tower\diskx (where x = the drive you cleared) and see what it shows.    I'm surprised if Joe's utility SAYS it cleared a drive, but didn't.    I suspect one of two things happened:

 

(1)  It actually cleared the drive, but meanwhile UnRAID was maintaining parity ... which is why it was taking so long.    If that's the case, the drive will still be in the array, but will be empty.

 

or

 

(2)  The drive was actually cleared "outside" of the array (i.e. UnRAID didn't "know" about it and wasn't maintaining parity).  If that's the case, you MAY be able to reconstruct the drive to its original contents by (a) Stopping the array and unassigning the drive;  (b) Starting the array so it shows the drive as "missing";  © Stopping the array and assigning the drive back to the same slot;  and (d) Starting the array and letting it rebuild the drive.

 

... OR you may have actually discovered a really weird thing about the Pre-Clear script 

[trurl]

Maybe you'd better post a screenshot of the main webGUI page. What makes you think you cleared an array drive, and what makes you think it still has data on it? If unRAID took it out of the array because it thought it failed then it would still seem like it still had data on it, but that would just be because unRAID was simulating the drive by using the other drives and parity to reconstruct the data. See this wiki which says:

 

Do not be misled by the fact that you can still read and write to the drive with a red ball indicator.

[garycase]

If the drive was cleared "outside" of the array, but never written to "in" the array, then there wouldn't have been any write attempts -- so UnRAID wouldn't have red-balled it.

 

Further, you DO NOT want to write to it -- because the write would likely be successful, but would result in an update to parity which would make it impossible to restore it to the original contents !!

 

If you just READ from the disk share (as I suggested above), the read should be successful -- but will likely show that there's nothing on the drive.    If you then Stop, unassign the drive, and Start the array so the drive shows as missing, you could then attempt to read from it [\\Tower\diskx] and it would show you what WAS on the drive, as it would be simulated via parity and the other disks.    If that happens -- then what I said above will work just fine ... Stop, reassign the drive, and then Start the array and let it rebuild the drive.

 

But do NOT do ANY writes to the drive or this won't be possible.

 

[vjmcdonnell]

I have gone in directly to disk 1 & 2 and played movies from them so data is still there.

I've discovered something strange then. :-)

 

[garycase]

I have gone in directly to disk 1 & 2 and played movies from them so data is still there.

I've discovered something strange then. :-)

 

Which disk was it that you cleared?    Post a screen shot of your Web GUI.

 

[vjmcdonnell]

Was disk 1 but checked and everything is there ad working.

 

[trurl]

I don't remember any mention of the fact that you don't have a parity drive. I guess that makes the whole discussion pointless then.

[garycase]

True.  With no parity drive, all discussion of possible ways to recover was clearly wasted.

 

In addition, while it's certainly a good idea to run a pre-clear to test the integrity of a new drive, it is NOT necessary to clear new drives to add them to the array => they don't require clearing if the array isn't parity protected.

 

It is also VERY strange that the data was still on the drive after you pre-cleared it.  Are you CERTAIN it was the same drive?      It'd be VERY interesting to see just what was recorded in the preclear report for the drive -- please post it here.  [it is on the flash drive in the "preclear_reports" folder.

 

[vjmcdonnell]

I will post the preclear report when I get back hope, 6+ hrs.

I just like the idea of running preclear before adding to the array.

 

Deffo the wrong drive as only have two 3tb drives.

[itimpi]

The only way I could see data surviving a pre-clear was if the pre-clear before the pre-read phase completed.

[Joe L.]

The preclear screen shot said you cleared /dev/sdd, but that disk on the WEB-GUI is disk2, not disk1, and it is a 2TB drive, not the 3TB drive you cleared.

 

If your data is still on both data disks, consider yourself very lucky.  The only thing I can think of is you either hot-plugged a drive into the array, or rebooted between those two screen shots, and the drive /dev/sdX designatoins were re-assigned..

 

Please post the preclear reports as requested, they will give the model/serial number of the drive you pre-cleared.

 

Remember, the dev/sdX designations can change EVERY TIME you reboot. They are assigned as the OS recognizes the disks after they initialize.  That does not always occur in the same order,especially with the same make/model drives.  You may have cleared the new 3TB you intended drive after all.

[vjmcdonnell]

I'm preclear in the correct drive at the moment and 12 hrs in and nearly finished.

 

I am so lucky I didn't lose anything

[garycase]

You may have cleared the new 3TB you intended drive after all.

 

That's why I asked to see the pre-clear report ... it's almost got to be the case.  I can't imagine that the pre-clear status would show the drive was cleared when in fact all the data was still on it !!

 

Doesn't explain why it was running so slow ... but at least it will explain the end result 

[vjmcdonnell]

All preclear reports

pre_clear.zip

[garycase]

Well, the 3TB disk in your array is NOT the same disk you were pre-clearing in your first post.

 

The only time WMC4N0848204 was pre-cleared was on 21 Feb, and when it was pre-cleared it was disk sdb.  The disk you show in the first post was sdd.

 

So that explains why the data is still on the drive in your array 

 

I suspect you may have been pre-clearing the disk you wanted to; and for some reason it was running very slow -- but when restarted, it worked okay.

 

[vjmcdonnell]

When I restarted it I remounted in the server,  HP microserver,  and this seems to have fixed the issue.

 

Thanks for all the help.