This night disk1 of my array showed lots of errors during a non-correctional parity check but parity check reported no sync errors. I do not understand that.

 

Please have a look at the attached screenshots and diagnostics. Even after completion parity check reports no errors while the read errors became more and more.

 

A short SMART test did show no error.

 

Two questions:

 

* If a drive has errors how can a parity check be successful?

 

* What is your suggestion regarding that disk? Cabeling is ok, adapter is ok, this drive never showed errors before this nights parity check.

 

Many thanks in advance.

tower2-diagnostics-20160909-0119.zip

So from my understanding the parity is fine. The error you're getting is when a disc has bad sectors or some other error that may alert that the disc is in the early stages of failure. If you go to the S.M.A.R.T. test for that disk, you should see some errors...mostly likely with bad blocks.

Thanks for your answer.

 

What I don't understand: If there are sectors that could not be read on disk1, and unRAID couldn't read them, how can parity check report successful completion?

 

I don't get that.

 

When there's a read error unRAID uses all other disks plus parity to calculate that data, it then attempts to write it back to the problem disk, if the write fails the disk is disabled, if it succeeds only the read error is logged.

 

SMART for that disk looks fine, but that doesn't mean there's not a problem, what I usually do in these situations is use the disk on another server (or trade cables/enclosure with another disk on the same server) and if the same disk has issues again replace it.

Thanks for your answer.

 

So for these ~2000 blocks parity could not be checked and a "rebuild" took place.

 

Hmm, interesting.

 

Sorry, need to clarify:

 

If I issue a non-correcting parity check on a dual-parity 6.2 rig and "Errors (for a disk) > 0" but "Sync errors detected = 0". Parity is still correct? Really?

 

TIA

 

Yes