6.9.2 New SuperMicro server Upgrade Drive Issues

[Flendor]

Please forgive me if this is in the incorrect place. I have searched for something similar and have struggled to find anything that helped.

 

I am migrating from a Fractal Design 7XL to a SuperMicro 846 rack due to needing to expand the volume of drives. (wish I had seen the 3D printed 24 bay option before I started this)

Supermicro X9DRi-LN4+/X9DR3-LN4+, Version REV:1.20A
American Megatrends Inc., Version 3.4
BIOS dated: Wed 20 Nov 2019 12:00:00 AM GMT

Adaptec ASR-71605 controller with one cable going to each of the front and rear planes

24 front bays, 12 rear bays

Two Intel® Xeon® CPU E5-2680 0 @ 2.70GHz

96 Gb DDR3

 

First attempt at moving the drives over, random drives would become disabled and error counts would go through the roof

Second attempt, parity rebuild would have the same impact, drives disabled and errors on all drives

Third attempt, Preclear would have the same impact, drives disabled and errors on all drives

 

Put everything back in the original build, all drives are fine, parity rebuild is fine, preclear new drives is fine.

 

Built the new server with a bunch of small size drives that I could beg and borrow to try and work on the issue without losing my data.

Preclear these drives causes the same issue. It can maybe cope with one 1TB clear before things start to fail.

Formatting the first array start causes the same issue.

Parity rebuild again has the same issue.

 

I have attached the log that I could download before things got hairy and loud noises came from the machine. Really have very little idea what I am looking at in these logs, other than red means bad...

 

Your help and assistance would be very much appreciated.

 

syslog-2.txt

[JorgeB]

Looks like a controller problem:

 

Aug 17 11:12:39 Malakai kernel: aacraid: Host bus reset request. SCSI hang ?

 

If available try with a different HBA, ideally an LSI.

[Flendor]

21 minutes ago, JorgeB said:

Looks like a controller problem:

 

Aug 17 11:12:39 Malakai kernel: aacraid: Host bus reset request. SCSI hang ?

 

If available try with a different HBA, ideally an LSI.

You mean replace the Adaptec ASR-71605 - is that the HBA?

[JorgeB]

Yep.

[JorgeB]

You can also try it in a different slot to see if there's any difference.

[Flendor]

1 minute ago, JorgeB said:

Yep.

Thanks, does it matter how many ports?

 

1 minute ago, JorgeB said:

You can also try it in a different slot to see if there's any difference.

Tried this to no success.

[JorgeB]

Just now, Flendor said:

Thanks, does it matter how many ports?

Just need one, though most will have at least two, and using two you'll have more bandwidth, but not really needed for 10 disks.

[Flendor]

7 minutes ago, JorgeB said:

Just need one, though most will have at least two, and using two you'll have more bandwidth, but not really needed for 10 disks.

the aim is to have most of the 36 bays populated at some point

[JorgeB]

With 12 disks or more you should use dual link.

[Flendor]

1 hour ago, JorgeB said:

With 12 disks or more you should use dual link.

Thanks, the front back plane of 24 bays only has one connection for the HBA, the other two connections appear to be for connection to extra back planes

[JorgeB]

23 minutes ago, Flendor said:

Thanks, the front back plane of 24 bays only has one connection for the HBA

Unlikely i it's a Supermicro backplane, what model is it?