Unraid version: 6.12.3

Server: Dell R720

Disks: 8 x  4TB / Mixed brand / Single disks each in raid 0 on Dell Perc controller

Problem:

 

* Disk 5 failed, purchased a new 4TB disk and attempted array rebuild - it reached 100% and the speed went from 50mb/s to around 10 k/bs - left for a few days but it didn't complete

* Rebooted, attempted again, reached 100% but wouldn't complete as before

* My bigger plan is to move to a new server and to a proper controller config (passing through the disks directly), so I purchased 3 x 18TB disks and installed a H200 HBA in IT mode.  The 18TB disks show up as expected...

* I rebuilt the array using one of the 18TB disks as parity 2.  This also had the same issue of getting to 100% but wouldn't complete

* Reboot again, this time attempted to replace the faulty disk 5 with the 4TB disk I purchased to fix the array, but I put the array in maintenance mode - As of writing this post, this is currently at 100% and still running at around 400kb/s (and getting slower) - the diagnostics file (below) have just been created in this status.

Notes:
*I have had this server for 3+ years and had a couple of failed disks in that time.  The array has always rebuilt as expected, though this was a previous version of unraid.
* I reseated the HBA, HBA cables and reseated the disks
*The long term plan is to add the 3 x 18TB disks to the array on the H200 (in IT mode), remove the 4TB disks from the array, then move the H200 and 3 x 18TB disks to a new server  (this time without the constraints of a PERC controller with multiple raid 0 configurations!)  If there is another way to migrate from one server to another, retain my unraid configuration and transfer the data to the new 18TB disks, then that could be an option. One thought was to create a ZFS pool, move the data, then migrate to new hardware, but I don't really want a separate pool, I'd rather use the normal method.

 

unraid-diagnostics-20230814-1357.zip

I've seen this before multiple times with PERC controllers, do you remember the last release it worked correctly?

Hi Jorge

 

Interesting. Yes, it definitely worked on a previous version of Unraid with the PERC controller, probably 6.11.5 (possibly January this year).

 

Are there any risks or conflicts if I restore to the previous version of 6.11.5?

 

Thanks

Simon

 

 

Should be fine as long as you are not using zfs, also see this:

https://docs.unraid.net/unraid-os/release-notes/6.12.0#upgrade-notes

So i downgraded to 6.11.5 and then attempted the rebuild, but it's done the same thing, saying it's reached 100% and slowed down to arrive 50 kb/s.

 

At this point it's likely a problem with the perc controller and i have to think of another way to migrate.

 

Remembering i have a new (to me) H200 HBA and 3 x new 18tb drives, if i transfer the data from the array to a spare nas, what will happen if i move the unraid usb stick, cache drives WITHOUT the existing array? In other words, can i move the unraid usb stick to a new server with the hba and 3 x disks, , create a new array, recreate existing folders and shares, then transfer from a backup?

 

Thanks in advance

Pretty sure this was already a problem with 6.11.5 (if it's Unraid release related).

 

10 hours ago, smickers said:

In other words, can i move the unraid usb stick to a new server with the hba and 3 x disks, , create a new array, recreate existing folders and shares, then transfer from a backup?

You can.

My issue [Array rebuild reaching 100% but never completing] turned out to be a partial faulty backplane in the Dell R720

 

As all my important data is backed up to the cloud, I bit the bullet and moved the Raid 0 drives from the PERC controller to the H200 HBA. I connected the H200 to the existing backplane, and boom, loads of smart errors, mostly CRC errors. I then connected the H200 directly to the disks and the CRC errors stopped.  FYI, the backplane is divided into two sections of 4 drives each, with one cable to each section; it was only the second section that had failed.

 

As expected, moving the drives from Raid 0 to a direct connection resulted in unmountable errors, so I followed the advice @JorgeB gave in other posts to rebuild each drive. Eventually everything rebuilt fine.  I'm now in a new Dell server with the H200 and currently running unbalance in order to move to the 18TB disks.

 

Two takeaways that may help others.
 

1. it's been mentioned several times and ignored by many (including myself) - don't do a raid 0 on each drive and pass through to Unraid - eventually this will catch up with you!  Seriously, don't do this. Unraid friendly HBAs are so cheap, just buy one.
2. Don't put your Dell server in the loft (in the UK anyway). Some of the casing - internal and external - has lightly oxidised, so I think the fluctuations in temperature and spells of humidity have upset some components. (It also has some fan redundancy errors that turned out to be the MB).
 

Thank you @JorgeB for your help here and your other posts.