Does anyone know how i can identify the drive causing the parity check errors? i get 5 every time. Is there a way to make Unraid remember the values on each disk at these error spots and then after a another parity check compare to see what drive is changing these spots? and have ability rerun at these spots to see if its fixed? i have 11 drivers many different sizes maybe i can get a notification when error happens so i can at least narrow it down if it gets passed some smaller drives without errors.

 

or am i limited to having to take a drive out one at a time with a new one rebuild the data and then check with a party check that takes like 20hours? each time?
 

Edited February 21, 20215 yr by EMKO

25 minutes ago, EMKO said:

i get 5 every time.

Are you doing a correcting parity check?

 

But, other than that Marvel based controllers will on (depending upon the drives which are installed) consistently cause 5 errors every time a parity check is done.  Marvel is not a recommended controller for this and other reasons.

yes it is doing a correction every time, i am using two SASLP-MV8 not sure if they are based on Marvel controllers, although i didn't get these errors for a long time.

https://forums.unraid.net/topic/59091-marvel-issues-starting-point-for-investigation/

 

At the end of the day, LSI based controllers are the gold-standard.  AsMedia controllers will also work good.  Marvel has lots of problems which you may or may not encounter (although you're hitting one of them now)

oh crap not what i was hoping for well maybe i can see if i can get this one drive onto my motherboard SATA

 

thanks

I used to use a SASLP-MV8 in my array, for about 7 years.  I would occasionally get 5 parity errors as you have done.  I replaced that controller with an LSI SAS2008 (PCI-Express Fusion-MPT SAS-2 [Falcon] (rev 03) ) and have had good service from it.  However, just yesterday I had a parity check complete, and I have 5 errors. 

 

Since this pattern has been related to the SAS controller, does anyone think I have something to worry about with this one?  I think I will run a second parity check.  In the past, when I did this, the same 5 errors would show up, and a third check would show no errors, as if they had been corrected.

 

Any ideas?

5 minutes ago, PeteAron said:

However, just yesterday I had a parity check complete, and I have 5 errors. 

If it is the first time with the LSI it's likely not controller related, still important to try and find the cause, any unclean shutdowns?

No, Jorge, It has been fine.  The one unusual issue i have had is an st800dm drive has been throwing crc errors- enough that i am going to return it.  I dont see others, so i assumed that isnt the issue.  

 

How do i find where the errors are?  Maybe its time to replace my 2 year old sas to sata cables?

Would writing to the array during a parity check possibly cause the errors?

Just now, PeteAron said:

Would writing to the array during a parity check possibly cause the errors?

Nope.

Well I am running a second parity check and, as expected, I see these 5 errors again.  Hopefully this means they are corrected; but is it possible they are scrambled?  I am writing corrections to parity; I dislike doing this because eventually I need to replace a disk, and i worry about the potential for these errors to corrupt my rebuild.

 

Is it possible this is a memory-related error?  I will replace my sas-sata breakout cables this weekend and run a third parity check.  Any input here is welcome.

10 hours ago, PeteAron said:

I see these 5 errors again.  Hopefully this means they are corrected

If the first check was non correct this is expected.

No, both checks were correcting.  I am going to replace my breakout cables and try again.

Update.  Last night i replaced both breakout cables from my LSI card to the disks, and started a new parity check.  I have the same 5 errors again. 

I am attaching a diagnostic in case anyone can identify something to help track down the issue.  I have the option of swapping hardware but obviously would not like to do that.

 

thanks for your help.

repository-diagnostics-20210413-2252.zip

19 hours ago, PeteAron said:

No, both checks were correcting. 

Diags posted show first a non correcting check, then a correcting one, so errors on the 2nd check were expected.

Jorge, sorry, I wasnt clear.  That log is from right before shutting down last night to replace the breakout cables.  I restarted and began a parity check - which is not in that diagnostic.  The currently running parity check again shows 5 errors, so they are being corrected a second time.  I will upload another diagnostic tonight when the check finishes. 

3 minutes ago, PeteAron said:

The currently running parity check again shows 5 errors

Are they the same 5 blocks? You're still using a Marvell controller for one disk, can you connect it to a different one?

They are the same five blocks.  I thought I eliminated the Marvell controller when I put the LSI card in to replace my SASLP?  

 

Do I have the wrong card now?

OK that answer was incomplete...Im slow.  This is a supermicro C7P67 with 6 onboard SATA controllers, and I have an LSI SAS2008 for 8 more.  All 8 of the LSI ports are used, and 5 of the onboard.  I see in the log where it says I have scsi host1 and scsi host 2 using Marvell - these must be on the motherboard, right?  how can i change that?  Why would only two have marvell controllers?

Yes, it's onboard, according to latest diags LSI only has 7 disks connected, connect the one using the Marvell there also.

Ah, yes, you are right - not enough coffee.  I will try that tonight.  

 

Is there no way to replace that marvell controller?  I suppose i need to avoid both of those, limiting my array to 12 disks?

Just now, PeteAron said:

Is there no way to replace that marvell controller?

Only if you replace the board, you can just not use it.

Crap!  Yes, I see that now - page 16 of the manual shows a marvell 88SE9128 controller for 1 PATA and 2 SATA-III ports.  Jesus, I had no idea.

 

 

thank you very much for identifying this for me, Jorge.  

 

https://www.supermicro.com/manuals/motherboard/P67/MNL-1279.pdf

OK last question, then.  Is it possible to use my cache drive on the marvell controller without issue?  I was planning to add a cache.

28 minutes ago, PeteAron said:

Is it possible to use my cache drive on the marvell controller without issue? 

Difficult to say for sure, Marvell controllers in general should be avoided with Unraid, though some are not so bad as others.