Read errors after upgrading two disks on 6.9.2

[Belgian Genius]

My setup was a Thecus N5550 with 6 drives installed:

 

1. Parity - WDC-WD121KRYZ - 12TB

2. Disk 1 - WDC-WD121KRYZ - 12TB

3. Disk 2 - HGST_HUH721212 - 12TB

4. Disk 3 - 3TB WDC Red

5. Disk 4 - 3TB WDC Red

6. Cache - Samsung SSD 850 Pro 256GB

 

I decided to upgrade Disks 3 and 4 to 10TB Ironwolf Pro.  There was no data stored on these drives, it just shows 69.8GB on each.

 

First I replaced Disk 3 with 10TB Ironwolf Pro.  Array rebuilt just fine and all was good.

 

Then, I replaced Disk 4 with 10TB Ironwolf Pro.  Array rebuild failed giving read errors on new disk.  So, I assumed new drive was bad, so I removed it and tried to return the original 3TB WDC Red.  Unraid then refused to accept it, claiming new disk needs to be larger than previous disk and expecting at least 10TB.  So, then I got a new 10TB Seagate Barracuda Pro and put it in Disk 4.  This rebuild also failed giving read errors.

 

Surely both of these new drives are not defective, so what gives

unraid-diagnostics-20210416-1431.zip

Edited April 20, 2021 by Belgian Genius
Add diagnostics.

[John_M]

Probably a bad SATA cable but without diagnostics all anyone can do is guess.

 

[Belgian Genius]

What kind of diagnostics should I post

 

There are no cables in this Thecus NAS.  It is PCB.

Edited April 20, 2021 by Belgian Genius

[John_M]

Tools -> Diagnostics. Post the zip file.

 

Cables are just connectors joined by wires. The PCB has connectors. It's the connectors that cause the problems because of their mechanical nature.

 

[Belgian Genius]

OK, obviously I am new here.

 

I added the diagnostics I ran after the first rebuild fail.

[JorgeB]

Apr 16 02:31:15 UnRAID kernel: md: disk4 write error, sector=5920974112
Apr 16 02:31:15 UnRAID kernel: md: disk3 read error, sector=5920976928
Apr 16 02:31:15 UnRAID kernel: md: disk0 read error, sector=5920976936

 

Multiple disk errors suggest a controller/power problem, since the controller is Intel I would start with power.

[Belgian Genius]

Interesting.  When drive 4 is offline, the parity check passes just fine.

 

I wonder if the power supply in this Thecus unit is fading out...

[Belgian Genius]

Well, I owe you a big thank you for this one!

 

The Thecus N5550 I had my disks in was at least 8 years old, so I decided to chuck it.  I have lots of parts laying around, so I constructed a server based on:

 

Asus X99 motherboard

Xeon e5-1620 CPU

16 GB Ram

...and all my disks.

 

I migrated everything to the new machine.  I poked around for a bit trying to get the machine to boot from the USB key, but when it did, it fired right up, automatically recognized all my disks, and presented everything to me exactly as it had been set up on the Thecus.

 

I instructed it to add Disk 4 to the mix (using the Seagate Iron Wolf Pro I had first tried), and 12 hours later it's all good!

 

Unbelievable.  I'm sold on UnRAID forever.