168 read errors-smart overall health passed

[baracas]

So I set up my unraid server over the last week or 2.  Everything went well.  Finally got around to running a read check (no parity drive yet) and got 168 errors on one drive.  There umc errors I believe, Diagnostic attached, but still.  I had funds and "planed" on adding a parity drive this month, but given the status of that disk, I'm guessing the best course of action would be to move some data around so that I can remove that data disk and replace it.  And just have to add a parity drive at a later time when funds become available again.  Or if there's something else I could do, I'm open to suggestions as what to do.  I haven't tried replacing the sata cable yet, but I will.  But I'm doubting thats the issue, its an old drive.  Thanks.

p67-diagnostics-20200706-1515.zip

[JorgeB]

10 hours ago, baracas said:

and got 168 errors on one drive

Since the diags are after rebooting you can't see which drive and what type of errors, I assume disk3 since it appears to be failing, you can run and extended SMART test to config, then copy all the data you can or use ddrescue.

[baracas]

Yeh, sorry I didn't specify its disk 3.  I'm in the process of moving data off of it to other places.  Extended smart tests error out.  I had heard that read errors weren't necessarily the end of the world.  But if I were to put parity disk in the array with a disk getting read errors like that, what would happen during a correcting parity check?  I don't exactly know myself.  But just seems, with my limited knowledge, unwise to leave a suspect disk in a array for long term.  Maybe after zeroing it out or something, it may still have a home as a unassigned device or something.  idk.  Thanks for responding.

[JorgeB]

31 minutes ago, baracas said:

But if I were to put parity disk in the array with a disk getting read errors like that, what would happen during a correcting parity check? 

Now it wouldn't be able to sync 100% because of the read errors.

 

32 minutes ago, baracas said:

But just seems, with my limited knowledge, unwise to leave a suspect disk in a array for long term. 

Correct, parity needs all others disks except one to recover, so a known bad disk should be replaced ASAP or if another one fails you'll be in trouble.