11 minutes ago, JoeUnraidUser said:

when hard drives have failed, I was still able to copy the emulated data off to other drives.

Then what? You still have to rebuild.

18 minutes ago, JoeUnraidUser said:

Physically I am replacing the 4 3TB drives with the 4 14 TB drives.  I am going to then make 1 of the 14TB drives the parity drive and 3 of the 14TB array drives and make the original 6TB parity drive an array drive.

Let me see if I can say this in a way that makes more sense to me and see if you agree.

 

You want to replace 4 disks of various sizes with 4x14TB drives. One of the 14TB drives will replace 6TB parity. 3x14TB drives will replace 3x3TB data drives. Then original 6TB parity will replace 1x3TB data drive.

 

21 minutes ago, JoeUnraidUser said:

So, no matter what happens I am going to have to rebuild parity. 

Obviously current parity isn't large enough to allow 14TB drives in the array.

2 minutes ago, trurl said:

Then what? You still have to rebuild.

I just replaced the bad drive with the new one and there was no need to rebuild since I had moved all the data off the emulated drive.  I just ran a parity check and moved the data back over from the other drives.

Just now, JoeUnraidUser said:

I just replaced the bad drive with the new one and there was no need to rebuild since I had moved all the data off the emulated drive.  I just ran a parity check and moved the data back over from the other drives.

Not enough detail in that description. At least, what little you did say there is not possible.

 

Parity would not be valid if you did it that way. And it wouldn't let you replace the disk without rebuilding anyway. Did you New Config and rebuild parity?

5 minutes ago, JoeUnraidUser said:

replaced the bad drive with the new one and there was no need to rebuild

Absolutely a need to rebuild, either rebuild parity or rebuild the replaced disk. No way around that and still have valid parity.

1 minute ago, trurl said:

Not enough detail in that description. At least, what little you did say there is not possible.

 

Parity would not be valid if you did it that way. And it wouldn't let you replace the disk without rebuilding anyway. Did you New Config and rebuild parity?

I can't remember but I must have done a New Config to remove the bad drive from that disk assignment and assigned the new disk to that disk assignment and I do remember doing a parity check after adding the disk.

I suggested the parity swap because he mentioned the 12/12 case capacity limit.  Take out a 3, parity swap the 14 for the 6, which adds the 6 to the array.  Not sure why that isn't the correct advice.

 

1 minute ago, JoeUnraidUser said:

I can't remember but I must have done a New Config to remove the bad drive from that disk assignment and assigned the new disk to that disk assignment and I do remember doing a parity check after adding the disk.

It would have done a parity sync (rebuild) by default when you did New Config unless you checked the parity valid box (and it wouldn't have been valid).

 

Rebuilding disks is what parity is for. I have upsized each of my array disks many times over the years by rebuilding. Never considered copying the data anywhere, seldom had any problem, and when I did I corrected it (usually a bad connection).

31 minutes ago, JoeUnraidUser said:

after hours of waiting it got all the way near the end and failed

And usually you can see whether it is having a problem without waiting until the end. And even when you don't notice it is having a problem, you can still correct it and do it again since it is still emulated.

 

3 minutes ago, musclecups said:

I suggested the parity swap because he mentioned the 12/12 case capacity limit.  Take out a 3, parity swap the 14 for the 6, which adds the 6 to the array.  Not sure why that isn't the correct advice.

Not necessarily wrong, just not necessary.

 

If you make one of the 3TB a missing disk, then you can do the parity swap as you suggest.

 

But the way parity swap works is by copying parity to the new disk, then rebuilding the missing disk to the original parity disk. And while parity is being copied, nothing can be allowed to change parity, so the array is offline.

 

So, little or no time saved since new parity has to be completely written, and then the replaced disk has to be rebuilt. And the array has to be offline while parity is copying.

8 minutes ago, musclecups said:

because he mentioned the 12/12 case capacity limit.

Normal disk replacement doesn't require any additional slots.

3 minutes ago, JoeUnraidUser said:

backup the data

  

48 minutes ago, trurl said:

Do you have another copy of anything important and irreplaceable? Parity is never a substitute for backup.

 

23 minutes ago, trurl said:

Absolutely a need to rebuild, either rebuild parity or rebuild the replaced disk. No way around that and still have valid parity.

 

See if this wiki makes sense to you and helps you understand why rebuild would have been needed.

 

https://wiki.unraid.net/Manual/Overview#Parity-Protected_Array

 

Understanding parity is the key to understanding how Unraid works, why it makes you work with the array the way it does, and how you must work with it.

1 minute ago, trurl said:

this wiki makes sense

Notice something that isn't mentioned at all there in that section about parity is "files". It is all bits. Parity knows nothing at all about files. Everything you do with copying, etc. is changing those bits. Normally when you change any bits on disks in the array, parity is updated along with those changing bits. But when you take another disk with files on it (or even no files on it) and put it in the array in place of another disk like you said you did, then none of those bits is in sync with parity.

  

17 minutes ago, JoeUnraidUser said:

Since getting burned by it twice many years ago, I have no faith in it anymore

Reading an emulated disk is just reading all the other disks to get its data from the parity calculation. Rebuilding a disk is just reading all the other disks to get its data from the parity calculation, then writing that data to the replaced disk.

 

Almost certainly any problem you had before was connections you had disturbed when replacing a disk. Often problems can be seen in Main - Array Devices in the Errors column and in syslog without waiting for rebuild to get all the way to the end and then fail. Most likely, it didn't even get all the way to the end, it just stopped because it failed due to bad connections.

On 1/9/2023 at 3:14 AM, JoeUnraidUser said:

Do you have to do a preclear on a brand new hard drive that you add to the server? 

No. It's your choice to test your drives before putting them in your server. Exactly as it should be.

 

MrGrey.

 

On 1/9/2023 at 1:37 PM, musclecups said:

They're made for datacenters and weren't abused.  With a supposed MBTF of 2.5 million hours, I think they'll probably perform pretty well for me.  

And how can one be sure that they weren't abused (mistreated, neglected, mishandled etc.)?