[SOLVED] Read errors (4) on newly installed drive

[BradJ]

I added a 6TB (Disk 8 ) drive to my array a few days ago.  Woke up this morning to a notification of 4 read errors.  I do not currently see any SMART errors but I am running an extended check to be sure.

 

Reading the Wiki, it says to run a parity check as soon as possible.  Should this be a correcting or non-correcting parity check?  A previous topic on this said to make sure it is non-correcting but I want to be sure before I start this 26 hour process. 

 

Any other words of advice? Thanks.

 

tower-diagnostics-20201005-0953.zip

[trurl]

SMART for disk8 looks OK. I always do an extended test in this situation too, then just reset the error count if everything is OK (Main - Array Operation - Clear Stats).

 

When was your last parity check? Did it have exactly ZERO (the only acceptable result) sync errors?

[BradJ]

I run a monthly parity check.  The new drive has not been thru a parity check since I just added it a few days ago.  Yes, all of my previous parity checks have had 0 errors.

 

I will start a parity check as soon as the extended smart check is finished.  According to the Wiki a (successful) parity check should zero out the errors as well.  I'm still not sure about the parity check being a correcting one or not.

 

Edited October 5, 2020 by BradJ
spelling

[trurl]

2 minutes ago, BradJ said:

new drive has not been thru a parity check since I just added it

I always recommend a parity check any time there has been any change to disk assignments, whether add, remove, or replace, just to verify everything is working as expected.

[BradJ]

Ok, that's good advice and I will run a parity check asap.  

 

Again, any insight if the parity check should be a correcting one or not?

 

[itimpi]

45 minutes ago, BradJ said:

According to the Wiki a (successful) parity check should zero out the errors as well.  I'm still not sure about the parity check being a correcting one or not.

 

That would only be true if it is a correcting check.   With a non-correcting check any parity errors are NOT corrected and would continue to be reported until a correcting check is run.

[BradJ]

 

7 minutes ago, itimpi said:

That would only be true if it is a correcting check.   With a non-correcting check any parity errors are NOT corrected and would continue to be reported until a correcting check is run.

That would make sense.  The Wiki doesn't specifically say the errors will be zeroed out.  I guess I made that conclusion in my head after I was reading about how READ errors are fixed automatically.  I guess I just assumed the next parity check would zero out the errors.  

 

I will run a correcting parity check and report back (in 26+ hours).

 

[itimpi]

5 minutes ago, BradJ said:

I will run a correcting parity check and report back (in 26+ hours).

You only want to do this if you think the disk is OK.  Running a correcting parity check if a disk is giving read errors can cause spurious errors and end up corrupting parity.

 

[BradJ]

3 minutes ago, itimpi said:

Running a correcting parity check if a disk is giving read errors can cause spurious errors and end up corrupting parity.

 

This is why I asked about this in the OP.  It's a new drive (to this array) so I can not be 100% certain on the quality of this drive.  Alright, just to be safe, I will run a non-correcting check first and see what happens.  I'll report back on the results.

 

[itimpi]

1 minute ago, BradJ said:

This is why I asked about this in the OP.  It's a new drive (to this array) so I can not be 100% certain on the quality of this drive.  Alright, just to be safe, I will run a non-correcting check first and see what happens.  I'll report back on the results.

 

If you keep getting the same sectors reported in the syslog as having errors then this is a sign that the drives are reading OK.

[BradJ]

Update:  Suspect Drive passed the SMART extended test.  I then started a non-correcting parity check - successful with no errors.

 

Then it got interesting.  I started a correcting parity check and immediately got 134 read errors.  It is my belief something is just wrong with the drive. What exactly - who knows - but something is wrong with it.

 

Where to go from here is the question now.  I'm thinking of just cancelling the current parity check and replacing the drive.  Is that a good game plan or is there anything else I should consider?   

tower-diagnostics-20201007-1320.zip

[JorgeB]

It's logged as a disk problem, and since it looks like it's a white label drive probably a good idea to replace it.

[BradJ]

39 minutes ago, JorgeB said:

It's logged as a disk problem, and since it looks like it's a white label drive probably a good idea to replace it.

Lol, yes - It's a while label drive purchased from GoHardDrive.  This is the second WL drive from there that unraid has found problems with.   

I'll be replacing it with a 10TB WD shucked drive (yes, I'm cheap).... I'll report back once the array is rebuilt and a parity check has completed successfully.

   

Edited October 7, 2020 by BradJ
grammer

[BradJ]

I have now replaced the drive giving read errors and everything seems to be working correctly again.

 

Moral of the story: do not buy white label hard drives from goharddrive (at least for a NAS).  In total, I have had problems with 3 out of 4 WL drives.  One was replaced under warranty with an enterprise level WL drive and that one works fine.  The other ones started having weird intermittent errors right after the one year warranty.  

 

Thank you to all that has helped me thru this, I appreciate it.  I will now mark this topic as solved.