Can't rebuild parity

[borange]

Hi there,

 

I have a bad disks in my setup but I'm having trouble replacing it because one of my parity disks is disabled.

I've tried to rebuild it a few times but it always fails.

 

I have 2 'disabled' disks. Parity and Disk 2. I've tried just rebuilding just the parity disk and also tried to bring them back both at the same time. 

 

I know have a lot of disks with SMART errors. I'd like to start replacing them (starting with Disk 14) but I'm having trouble doing that.

 

Is there anything I can do to get the Parity back up? Or am I in trouble? (I'm assuming its failing the rebuild because of the bad disks in the array)

 

Specs:

Intel Xeon 6230R CPU 

Supermicro X11SPi-TF

80GB RAM

ADAPTEC ASR-72405 (2274900-R)

UNRAID 6.9.2 

 

dump-diagnostics-20210914-0753.zip

[trurl]

Most of your disks have disconnected. Are you sure you don't have a power issue?

[borange]

I think its good? Is there some form of test I can run? I've got a 1000Watt corsair PSU.

 

it been running fine for days/weeks - just Parity and disk 2 are disabled. Only seem to be running into problems when trying to rebuild the parity.
I guess all disks might be spinning during that process? 

 

 

[trurl]

All disks are read when checking parity or when rebuilding any disk or when emulating a disabled disk.

[borange]

I just ran a small stress test (loading up all 52 cores/thread with prime95) to see if anything happens. Figured this would be a good way to push power limits.

 

No problems with at all. I've got power adapter hooked up it as well and I didn't see that get over 500Watts (not a great measurement but better than nothing)

 

Not sure how else I can push it...

[ChatNoir]

How is power delivered to your drives ?

How many drives per lead from the PS ?

[JorgeB]

Looks like a controller problem:

 

Sep 14 04:40:04 dump kernel: aacraid: Host bus reset request. SCSI hang ?
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Adapter health - -3
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: midlevel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: lowlevel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: error handler-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: firmware-19
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: kernel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Controller reset type is 3
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Issuing IOP reset

 

Make sure it's well seated or try a different PCIe slot if available, if issues persist I would try a different controller, LSI recommended.

[borange]

12 hours ago, ChatNoir said:

How is power delivered to your drives ?

How many drives per lead from the PS ?

 

I think its via molex - 12 molex connectors on 4 different cables 

[borange]

11 hours ago, JorgeB said:

Looks like a controller problem:

 

Sep 14 04:40:04 dump kernel: aacraid: Host bus reset request. SCSI hang ?
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Adapter health - -3
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: midlevel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: lowlevel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: error handler-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: firmware-19
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: outstanding cmd: kernel-0
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Controller reset type is 3
Sep 14 04:40:04 dump kernel: aacraid 0000:17:00.0: Issuing IOP reset

 

Make sure it's well seated or try a different PCIe slot if available, if issues persist I would try a different controller, LSI recommended.

 

 

Will move it to a different spot and see if things are different. 

[borange]

I'm very surprised - Moving slots seems to have worked.

 

parity rebuilt! 

 

Total size:	10 TB	
Elapsed time:	23 hours, 55 minutes	
Current position:	10.0 TB (100.0 %)	
Estimated speed:	115.5 MB/sec	
Estimated finish:	completed

 

Thank you all for the help!