Unexplained disk errors when adding parity disk

[ZipsServer]

Just realizing I have been an unraid user for 10 years now.....

@trurl I just looked back at my Newegg history..... It is a Rosewill 550W 80+ Gold purchased back in 2012, along with the mobo and other components.... might be time to replace some parts?

 

@kizer No, only using the standard power cables that came with the PSU. The drive cage could be suspect. I am using the ICY DOCK MB974SP (also purchased in 2012)

[ZipsServer]

Update: I am now having disk errors with both disk2 and disk3. The log said they had bad super blocks and were unmountable. So the issue now exists even with the parity drive completely removed.

I just ordered a new LSI 9211-8i with SAS breakout cables, but it sounds like I should also purchase a new PSU? Any recommendations for powering 10-15 SATA drives? Most PSUs don't even to seem to come with cabling for that many drives...

EDIT: I may also accept recommendations on how to move away from those ICY Dock cages (if they are the problem) to something more quiet. I have this Xigmatek case

Edited September 3, 2022 by ZipsServer

[trurl]

25 minutes ago, ZipsServer said:

having disk errors with both disk2 and disk3. The log said they had bad super blocks and were unmountable.

Do you know how to proceed with that? After hardware is fixed of course.

[ZipsServer]

@trurl I actually just swapped drives around in the cage that seems to be problematic. Only disk2 is unmountable now. I am going to try to move disk2 out of that cage... maybe 2 or more slots in that cage are bad.... or maybe I need to buy some compressed air?

But to your original question, yes I think, if the actual superblock is bad then I can use fsck to restore one of the backups. However it seems like if the hardware problems is resolved then I might not have to worry about that. (Had similar issues with my external drives when the USB header was loose.)

mastertower-diagnostics-20220902-2225.zip

[trurl]

 

[JorgeB]

Disk2 appears to be failing to mount just because of the read errors, it should once those are solved.

[ZipsServer]

Hi, it's me again.

 

So I bought a new HBA (LSI 9211-8i), new SFF-8087 cables, and a new 650W PSU just to power the HDDs. All the new gear appears to work well.

 

However, the 10TB drives that were previously problematic are no longer appearing. They seem to be completely dead. I have tried them on both PSUs and both HBAs. Are there any other ways to confirm they are completely dead?

 

While waiting for the new gear to arrive I managed to transfer all the data from the 10TB disks to other disks in the array. I then removed them from the system and had them sitting on a shelf.

 

EDIT: Good news is that I am building parity with no issues.

mastertower-diagnostics-20220917-1641.zip

Edited September 17, 2022 by ZipsServer

[trurl]

1 hour ago, ZipsServer said:

the 10TB drives that were previously problematic are no longer appearing

Can you see them in BIOS?

 

1 hour ago, ZipsServer said:

new 650W PSU just to power the HDDs

Modular? Cables are not interchangeable and many people have fried disks trying to reuse modular cables from another PSU.

[JorgeB]

Also make sure it's not the 3.3v issue, google "SATA 3.3v pin" or power the disks(s) with a Molex to SATA adapter since that won't have 3.3v.

[ZipsServer]

Sorry for taking so long to respond. The parity build and then work got in the way.

 

I almost made the mistake of using non-matching PSU cables. Its kinda stupid the cables aren't made to the same specs...

 

The 10TB are not appearing at all. I can hear them spin up when I plug power in, but otherwise they are not identified in BIOS or /dev/sd*. I also tried powering both 10TB drives from the original power supply that they were working with (so not a 3.3V issue).

 

So for whatever reason, both 10TB drives started to error and then just died. I purchased these at the same time in 2020 and they are just a few months past their warranty.... These are my first drives to fail. I still have some 3TB REDs from 2013 doing 300k read/writes a day.