I think this is my first post here! The conversion to unRAID has been smooth. This is such a great community and so much documentation. I did see threads.on this I just wanted to make sure my plan was solid. I currently have an 8 disk array with 7D+1P. I am going to pull out at least 2 maybe 3 smaller disks and replace with 1 larger drive, so I'll have 5 or 6 data disks and 1 parity. All of my internal SATA ports are full.

 

I was thinking I can pull out one of the data drives to be replaced. Preclear the new drive, let it rebuild from parity. Then shut down dockers and mover then use unbalance to move data from the other drive I will remove. And then remove that drive from the array and run a parity check/write/rebuild... Profit?

 

I understand unbalance is pretty much just a gui for rsync but I've successfully used it in the past and I don't have to learn more command line lol

 

Sent from my Pixel 6 Pro using Tapatalk

 

 

What size are you actual drives (data and parity), what drives you plan to replace and what will be the new drive.

Your diagnostics could also help provide a better understanding of your system.

Parity drive is 14TB new data drive will be 14TB. Final result would be remove disk 6 (4tb) and 7(4TB), have new 14TB drive as drive 6. Depending how storage look maybe pull disk 5 as well.  If you need any more info please let me know! 

vulcan-diagnostics-20220105-0546.zip

I'm also going to run a parity check before doing all of this. My last one is only a month old or so but you can't be too safe. 

Your drives look OK, expect for disk7 that has some issues.

Those attributes are not considered by default by Unraid as they are not used the same way by every manufacturer.

The should be added manually for each WD drives.

 

  1 Raw_Read_Error_Rate     POSR-K   200   200   051    -    21
200 Multi_Zone_Error_Rate   ---R--   200   200   000    -    2

 

So I would start by replacing this drive first.

 

In your situation, I would :

• stop the server
• rebuild disk 7 on the new drive
• move the data around from drives you want to retire (either with Unbalance or manually, command line or with Midnight Commander)
• use the New Config tool to rebuild parity with only the drives you want to keep (we can discuss this later)

The preclear is optional but must be done outside of the Array.

 

Thank you sir. 

 

Every thread I've read says not to worry about "balancing" drives after something like this. Using high water any new data is going to be written to this new drive for quite a while so maybe it isn't as big of a deal but doesn't drive performance degrade as the drive fills? Wouldn't all the drives being 70% filled give better r/w speeds than them being 92% filled? 

36 minutes ago, GTvert90 said:

Wouldn't all the drives being 70% filled give better r/w speeds than them being 92% filled? 

Yes, especially for reads. For writes, the overhead from Parity makes it less true.

But it really depends on what is on the drives, how often it is accessed and what speed is acceptable.

 

Most people use Unraid to store data that is accessed very occasionally (media) and for which the speed is not a major issue.

If your use case is different, then this could be useful.

True. It's not often I'm performing a full backup/restore and a 50MBps speed hit when spinning up a 40mbps movie wouldn't even be noticed.

I don't think it would impact my usage but man all that red and orange bothers me haha. I might move one stuff around, I just don't want to cause more harm than good so maybe I won't.

Sent from my Pixel 6 Pro using Tapatalk

On 1/5/2022 at 7:15 AM, ChatNoir said:

Your drives look OK, expect for disk7 that has some issues.

Those attributes are not considered by default by Unraid as they are not used the same way by every manufacturer.

The should be added manually for each WD drives.

 

  1 Raw_Read_Error_Rate     POSR-K   200   200   051    -    21
200 Multi_Zone_Error_Rate   ---R--   200   200   000    -    2

 

So I would start by replacing this drive first.

 

In your situation, I would :

• stop the server
• rebuild disk 7 on the new drive
• move the data around from drives you want to retire (either with Unbalance or manually, command line or with Midnight Commander)
• use the New Config tool to rebuild parity with only the drives you want to keep (we can discuss this later)

The preclear is optional but must be done outside of the Array.

 

So the GUI is telling me that the new drive (14tb replacing a 4tb) is larger than the parity drive(Also 14TB).. Its equal but not bigger. Do I have to parity swap or am I missing something? 

Is one of those 14TB shucked from a USB enclosure ?

Your diagnostics with the new drive should provide more information.

3 minutes ago, ChatNoir said:

Is one of those 14TB shucked from a USB enclosure ?

Your diagnostics with the new drive should provide more information.

no, Both are x16 Exos 14TB drives vulcan-diagnostics-20220106-0738.zip

 

I don't mind parity swapping if that is what is needed. I just thought this was strange. 

Thank you for all your help. I couldn't really find the right keywords to find any posts on this issue. 

 

Disks have the same exact size, possibly the partition isn't starting on the default sector, post output of:

 

fdisk -l /dev/sdb

 

 

51 minutes ago, JorgeB said:

Disks have the same exact size, possibly the partition isn't starting on the default sector, post output of:

 

fdisk -l /dev/sdb

 

 

Disk /dev/sdb: 12.73 TiB, 14000519643136 bytes, 27344764928 sectors
Disk model: ST14000NM001G-2K
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disklabel type: gpt
Disk identifier: F81D82F2-BC67-4F1C-BA61-03CDE1B1ED89

Device     Start         End     Sectors  Size Type
/dev/sdb1   2048 27344764894 27344762847 12.7T Linux filesystem

Disk /dev/sdm: 12.73 TiB, 14000519643136 bytes, 27344764928 sectors
Disk model: ST14000NM001G-2K
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disklabel type: gpt
Disk identifier: 9A21185A-CEEA-4740-B6D8-B8F79ADEB5BB

Device     Start         End     Sectors  Size Type
/dev/sdm1     64 27344764894 27344764831 12.7T Linux filesystem

I ran the new disk for good measure too. 

10 minutes ago, GTvert90 said:

64

This is the Unraid default for disks, parity partition starts on sector 2048, it means it was partitioned outside Unraid, recently Unraid started supporting that starting sector for SSDs so it doesn't complain, but because of that there are a few less sectors available, to resolve this you'd need to wipe that partition then let Unraid repartition it and re-sync parity.

 

 

2 minutes ago, JorgeB said:

to resolve this you'd need to wipe that partition then let Unraid repartition it and re-sync parity.

But disk 7 is sketchy :

On 1/5/2022 at 2:15 PM, ChatNoir said:

  1 Raw_Read_Error_Rate     POSR-K   200   200   051    -    21
200 Multi_Zone_Error_Rate   ---R--   200   200   000    -    2

 

 

Do you have a way to backup it's content just in case ?

Yeah, I can back up disk 7 if needed. I actually already removed it as I tried to replace it. Can I perform a parity swap to resolve this? (Whatever that involves or Would I need to reinstall disk 7 for that? Whichever is the path of least resistance.

That's weird it is formatted like that. It's only ever been in an unRAID server. I can't imagine I formatted it anywhere else.

Sent from my Pixel 6 Pro using Tapatalk

Thanks for the help guys.

Sent from my Pixel 6 Pro using Tapatalk

3 hours ago, ChatNoir said:

But disk 7 is sketchy :

 

 

Do you have a way to backup it's content just in case ?

Would this be the procedure for parity swap on 6.9.2? I know its not "tested" but I'm good with an educated "assumption" in this case. I still have disk 7 to hopefully get data off if needed. This should work for me correct? https://wiki.unraid.net/The_parity_swap_procedure

Edited January 6, 20224 yr by GTvert90

I committed to the parity swap. Let's see how it goes

Sent from my Pixel 6 Pro using Tapatalk

Parity swap will also work.

Parity swap will also work.

So I think I messed up. Parity copied, but before I mounted and rebuilt data I thought I was going to cover my bases and format the old parity drive to avoid any issues in the future with the partition starting point. Now it's telling me config is invalid too manu missing disks. This isn't a huge deal. I can pop disk 7 back in and rebuild parity.. but is there an easier way?

Sent from my Pixel 6 Pro using Tapatalk

Do I want to do this? Can i do this with a replacement disk for drive 7 or would I want to put the original drive 7 back in and reset it with this? Will all the shares and other config like dockers and what not still be there?  Parity successfully copied but I made the mistake of formatting the old parity drive. In theory, if its possible I feel like I should be able to reset with the new formatted drive in disk 7 and everything should populate and rebuild properly. 

 

vulcan-diagnostics-20220107-1416.zip

So I put it back to the way it was (well with the drive the parity was copied to in the parity slot. ) and the original disk 7 in. I selected parity is valid (it should be) but I'm still running another parity check. If it completes successfully I'll swap out disk 7 and let it rebuild.

Sent from my Pixel 6 Pro using Tapatalk

If the parity copy completed successfully it would be, but not a bad idea to make sure.

And FYI for the future the parity swap procedure must be done completely without any interruption or it will abort and you'll need to start over from the beginning.