"md: disk3 read error" during parity rebuild

[hades]

Hello,

 

Long story, but I'm in the process of rebuilding parity. It's about 50% of the way done, but some errors are being found. The log shows this:

 

Jan 28 10:48:45 winserver kernel: md: disk3 read error, sector=2981773040

Jan 28 10:48:45 winserver kernel: md: disk3 read error, sector=2981773048

Jan 28 10:48:45 winserver kernel: md: disk3 read error, sector=2981773056

Jan 28 10:48:45 winserver kernel: md: disk3 read error, sector=2981773064

 

Total of ~290 so far. I am trying to interpret this. Some questions:

- Does this mean the disk is failing? Or cable perhaps?

- Is the data irreparably gone?

- Can I find out what files are affected?

- Should the disk be replaced?

 

The parity continues to build, the rate of building is consistent, and all drives are green.

 

Attached are the logs, and the SMART details from the webUI.

 

Any advice would be appreciated.

 

Thank you!

syslog.zip

[dgaschk]

Are you rebuilding parity onto a new disk or the existing disk?

[hades]

Thank you for the quick response.

 

I re-initialized the configuration, and set up the drives in the same order as before. So the parity was valid beforehand.

 

Thank you.

[dgaschk]

 

Attach SMART reports for all disks save cache (if any).

 

If the disks look ok then disk 3 can be rebuilt.

 

What was the reason for the rebuild?

[hades]

Hello,

 

The parity is being rebuilt because of a loose cable that put one of the drives in RED and things got completely haywire, I couldn't get the disk back to GREEN even after I confirmed that the cable was at fault and I replaced it. I decided to restart with a new configuration, and took the opportunity to add 2 drives to the array. At this moment I have a backup of the critical data, so I thought rebuilding the parity was the easiest choice.

 

However, these read errors are new. My usual stance is that if there are bad sectors on the drive, then I replace it. However, I'm not 100% sure how to do this check under unRAID, or how to interpret the drive being GREEN while showing these read errors?

 

Thanks.

 

[Squid]

The parity is being rebuilt because of a loose cable that put one of the drives in RED and things got completely haywire, I couldn't get the disk back to GREEN even after I confirmed that the cable was at fault and I replaced it. I decided to restart with a new configuration, and took the opportunity to add 2 drives to the array. At this moment I have a backup of the critical data, so I thought rebuilding the parity was the easiest choice.

Not that it matters now that you've done an new config, but for future reference, the reason why you couldn't get the drive back to being green is that when you have a red-balled drive and you want to rebuild back onto itself (ie - the cable was at fault), what you have to do is stop the array, from unRaid's menu switch the red-balled drive to be "not installed", start the array, then stop the array again and change the drive from uninstalled to the drive, and finally start the array.  It will then rebuild.

 

[lionelhutz]

The multiple changes to the array and the new config means you can't recover. With those bad sectors you won't be able to properly read the files from that part of the hard drive. I don't believe there is any way to figure out which files are affected.

 

It was a read attempt so the sectors are probably marked as pending reallocation which means you likely couldn't fully recover any failed drive at the moment.

 

[hades]

Thank you for the clarification. Still wondering:

- Does this mean the disk is failing? Or cable perhaps?

- Should the disk be replaced?

 

Thank you!

[lionelhutz]

The disk is reporting bad sectors, so it appears it's failing.