Read errors during parity check - what does Unraid do?

[hawihoney]

I'm currently running a parity check on a dual parity system. During that parity check one disk showed 128 read errors.

 

What puzzles me now is, why didn't Unraid throw that disk out? And what happens with the data that couldn't be read? Was ist replaced from the parity disks?

 

Is the data, that couldn't be read, still on that failing disk? Do the bits and bytes of both parity disks and the failing data disk still match or is there different data on these three disks now?

 

Diagnostics attached.

 

Thanks in advance.

 

tower-diagnostics-20210110-0856.zip

Edited January 10, 2021 by hawihoney

[JorgeB]

2 hours ago, hawihoney said:

What puzzles me now is, why didn't Unraid throw that disk out?

Because there weren't any write errors, if a read error is found Unraid uses parity and the other disks to calculate what should be there and re-writes that block to the offending disk, if there's a write error the disk will be disable.

[hawihoney]

2 hours ago, JorgeB said:

if a read error is found Unraid uses parity and the other disks to calculate what should be there and re-writes that block to the offending disk, if there's a write error the disk will be disable.

Ah, thanks. Something like that should be added to the logs IMHO. Currently the read error is written to syslog, the positive rewrite should be mentioned after that as well.

 

 

 

[JorgeB]

No need, no write error = successful rewrite.