Upgraded parity drive now can't finish a parity rebuild

[bmrowe]

I ordered a new 10TB WD drive (a shucked WD Elements external) and replaced my existing 8TB parity drive. I also moved that existing parity drive to become another data drive. However, building parity on the new hard drive has failed twice and not had a successful run yet. The first time, it made it to around 20% and the second time close to 95%. Both times, I've gotten warnings about read errors on the other drives and then the parity drive becomes disabled and I can't bring it back. The array is connected via USB-C and is a Mediasonic hf2-su31c.

 

The log for the drive right before failure shows:

Jan 27 15:42:13 Tower kernel: usb 2-4.1: Failed to set U1 timeout to 0x0,error code -19
Jan 27 15:42:13 Tower kernel: usb 2-4.1: Set SEL for device-initiated U1 failed.
Jan 27 15:42:13 Tower kernel: usb 2-4.1: Set SEL for device-initiated U2 failed.
Jan 27 15:42:13 Tower kernel: usb 2-4.1: usb_reset_and_verify_device Failed to disable LPM
Jan 27 15:42:13 Tower kernel: sd 1:0:0:0: [sdb] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x07 driverbyte=0x00 cmd_age=0s
Jan 27 15:42:13 Tower kernel: sd 1:0:0:0: [sdb] tag#0 CDB: opcode=0x8a 8a 00 00 00 00 04 50 f6 f3 48 00 00 08 00 00 00
Jan 27 15:42:13 Tower kernel: blk_update_request: I/O error, dev sdb, sector 18538230600 op 0x1:(WRITE) flags 0x4000 phys_seg 256 prio class 0
Jan 27 15:42:13 Tower kernel: md: disk0 write error, sector=18538230536

 

After the first time this happened, I reseated the drive thinking that perhaps that was the issue. And while it did get further, it did not complete.

 

Diagnostics attached. Ideas on what to try next?

tower-diagnostics-20220127-1751.zip

Edited January 28, 2022 by bmrowe

[trurl]

Looks like all of your disks disconnected. How are these attached?

[bmrowe]

2 minutes ago, trurl said:

Looks like all of your disks disconnected. How are these attached?

Its a small form factor pc connected to a mediasonic 4 bay enclosure via usb-c. 

[trurl]

USB connections are NOT recommended for array or pools for many reasons, including disconnects.

[trurl]

I usually say if you are determined to use USB for the array, forget about having parity. That way there is nothing to be out-of-sync when drives disconnect.

[trurl]

Also noticed corrupt docker.img, can't tell what disk it was supposed to be on since you have system share configured to be moved to the array and your cache disk is (mostly) empty.

[bmrowe]

Yikes. My HP ProDesk 400 G6 doesn't have a SATA port. Can I achieve the same with a serial port? Options look like: 

 

 

[bmrowe]

Or would an e-sata to usb cable get around this issue?

[trurl]

Best idea would be different computer.

[trurl]

You have a SATA controller, don't know how many ports. SATA to eSATA is just a different plug on one end of the cable. If you had an eSATA enclosure with separate ports for each disk, and there were enough SATA ports on the motherboard, you might make that work.

 

[trurl]

Some people apparently have made similar things work. Is there any configuration for that enclosure that might control whether it tries to save power by sleeping or something?

[bmrowe]

I don't think so. This is the manual: https://images-na.ssl-images-amazon.com/images/I/A1kRDk8X1iL.pdf It mentions: "USB port of your PC must support power-off function so that the device would go to sleeping mode. Setting up motherboard’s (power management ) in S3 is strongly recommended. For more details, please refer to user guide of motherboard BIOS setting."

 

I'm not sure if that is referencing the functionality it has called 'sync' where it sleeps when the pc sleeps. I have that turned off.