I manually ran the zero command on a drive in a tmux session:

 

dd bs=1M if=/dev/zero of=/dev/md6p1 status=progress

 

After several hours, I show it writing at ~5MB/sec. I have reconstruct write enabled. What am I missing here?

 

Edited April 25, 20251 yr by flyize

1. How is the drive physically connected to the server?

 

2. How busy is the array?

It's SATA and these are essentially the only writes happening right now.

Array reed speeds in the diags are very slow, try switching to normal write mode, if it speeds up, it could be slow reads on a different disk.

Seems to be about the same.

 

Edited April 25, 20251 yr by flyize

There are small reads from disk6 and parity, disk should not be happening during a clear, is anything else going on? If you there should be, try in maintenance mode, you can just start to see if it's same or not.

Nope, nothing going on. I assume I need to stop the zeroing so that I can move the array to maintenance mode? If its going to be this slow, I *really* don't want to have to start over, but certainly will if there is no other way.

6 hours ago, flyize said:

I assume I need to stop the zeroing so that I can move the array to maintenance mode?

Yes.

 

That speed is very far from normal, so something is going on, but cannot say so far if it's hardware or another issue.

I think I discovered the problem when I rebooted. Seems disk 13 died somehow?

 

Edited April 25, 20251 yr by flyize

SMART looks OK, run an extended SMART test.

So I ended up risking it for the biscuit. I set a new config, keeping everything, and the drive seems to be fine now.

 

I'm now back to zeroing out the drive listed above at about 200MB/s. I assume the zeroing process will rebuild parity as it goes, correct?

 

edit: Crap, it will only rebuild parity through 8TB. I guess I should fully rebuild parity first.

Hmmm, maybe it isn't okay

 

 

Edited April 25, 20251 yr by flyize

Disk dropped offline, this is typically a power/connection issue, you can swap cables/slot with a different disk, then try again, and see if the issue follows the disk.

Meaning likely a bad power or SATA cable?

Yes, that's what it looks like, swapping cables/slot should help confirm.

On 4/24/2025 at 12:24 PM, JorgeB said:

Yes, that's what it looks like, swapping cables/slot should help confirm.

So I swapped everything and it was stable for a day. Then it dropped off again. I started the array in maintenance mode, ran an xfs_repair to make sure nothing was wrong. It said that a log file needed to be replayed. I tried that and got another error (I don't recall). According to the message in Unraid, it didn't seem like I would lose any real data if I zeroed out the log - so I did. I'm now getting this error:

 

Unmountable: wrong or no file system

This seems bad. What have I done?

truffle-diagnostics-20250426-1444.zip

Post the output from xfs_repar on that disk.