New Drives Getting Read/Write Errors Randomly?

[Noraemsu]

Hello,

 

A few weeks ago I had major drive faliure where I think in the end I threw out 8 drives (4TB, 10TB drives). Did SMART and sector checks on all of them. Everything was good, but in unraid running preclear only one of the drives passed everything.

 

So I got some new 18TB drives and transfered all the data I had and the data I could recover, all in all pretty decent outcome all things considered.

 

The server was running fine for like 4 weeks or so, then one of the new 18TB drives got read/write errors (2048) of them I think.

I didn't think much of it, could've been a fluke. SMART data looks good, it's a brand new drive, passed the preclear.

So I did the rebuild on the same drive process. However not that long into the rebuild (in maintenance mode) another 18TB drive did the same thing read/write errors and also (2048) of them. So for now it says everything is emulated thanks to the dual parity drives but I still turned it off until maybe someone knows what's going on.

 

I swapped platforms back in november of 2022 and was running that fine until my first errors showed up in May (if not slightly earlier) this year. I remember with my other 18TB drives unraid had some issues getting them to work properly in the array so I had to tmp move them to a windows machine and use a software to alter the sector size or something I can't really remember, could that be the issue?

 

I tought my sas2008 had kicked the bucket so I ordered a new HBA. A Broadcom 9300-16i (LSI SAS3008 (rev 2)) but the issue still persisted.

 

I did a reinstall of unraid roughly a week ago when I also started getting flash drive corruption (which is not that suprising that flash drive has been with me since I started using unraid many years ago).

 

The only things I transfered over was my configs for docker containers and plugins.

 

I'm actually considering making this server a trueNAS box if this keeps up. At this point I'm at a loss.

 

dionysus-diagnostics-20230711-2343.zip

dionysus-drive5-smart-20230711-2357.zip

dionysus-drive6-smart-20230712-0003.zip

[JorgeB]

Disk dropped and reconnected, this is usually power/connection related, you should also update the HBA firmware to latest, since those are very recent disk models.

[Noraemsu]

The server is run using a Silverstone Strider 850w gold with 70A on the 12v rail and 22A on the 5v rail, should be fine?

 

The HBA was already the latest (or I couldn't find a newer one, did the FW and BIOS update anyways)

FW - 16.00.10.00

BIOS - 08.37.00.00

 

Since it already was the "latest" I'll revert back the PSU to the dual psu that came with the supermicro case.

[JorgeB]

15 minutes ago, Haagrid said:

16.00.10.00

That's no the latest, IIRC it's 16.00.12.00

[Noraemsu]

That's for the 9305-16i, I got 9300-16i. Can I flash the 9305 fw on a 9300?

[JorgeB]

So after doing a quick search there is a 16.00.12.00 firmware for the 9300-8i but it's to solve an issue with TrueNAS Core (FreeBSD not Linux), and it's not available on the Broadcom site, not sure if the 9305-16i firmware would work, but even if it did it likely wouldn't help so not much point, try the PSU

[Noraemsu]

I swapped to the old PSU's, so far everything is working. I'm not convinced yet and gonna do some more testing with the other PSU before I confirm that was the issue.

[Noraemsu]

After some more testing with the single ATX PSU everything seems to work after I changed the power cables to the backplane, instead of having one loom I now use two and so far no issues.