I done a bad thing...

 

I got an alert about my UPS being failed on my dashboard. It's 2-3 years old, so the batteries could be fried. So, I went down to investigate and was greeted by the UPS alarm. Unfortunately, my groping fingers found the emergency shutoff switch on the UPS. I was, at the time, writing some new rips to the array. I restarted the array and of course, it started doing a parity check because of the unclean shutdown. At the end of the check, I had 2727 parity errors reported. So, I set the write correct on and ran it again. I thought that this would return me to zero parity errors, but the system still reports 2727 parity errors. Oh, I am running on 6.3.2 right now. I think I updated to 6.3.2 before this started - it's been about 10 days.

 

Three questions:

  (1.) I don't doubt that the parity errors are real, but i expected them to be confined to one file and there are so many, it doesn't seem to fit what happened?

  (2.) What is the best way to return the parity to zero errors? I had not added any hardware for several weeks so it seems very odd that the errors would pop up and then remain exactly the same...if they were increasing, I would suspect a drive going bad but that has not happened? I think I simply have a corrupt file and everything else is OK.

  (3.) Is there any way to determine which files are corrupt?

 

I am attaching my downloaded files...if anyone has any suggestions?

tower-diagnostics-20170228-2240.zip

Edited March 1, 20179 yr by Unqualified Spectator

Did I miss something? You mention a noncorrecting and then a correcting, but then you didn't do another?

 

Haven't looked at your diagnostics yet. So after an unclean shutdown, you got a noncorrecting parity check (1st), which found errors. So you ran another correcting parity check (2nd), which found the same errors and corrected them. But you don't mention running another parity check (3rd) which should find zero errors after the 2nd parity check corrected the errors.

 

If you want it to say zero parity errors found on last check you have to run a parity check that produces zero errors.

 

I suspect the same thing trurl noted => you did a correcting check (which found and corrected the errors -- but it still reports how many it had found/corrected) ... but you didn't run another check afterwards ... which should then show zero errors.

 

Personally, I only run correcting checks with one exception -- after a drive rebuild (in the case I don't want parity changed if the rebuild didn't work perfectly, as I would want to redo the rebuild if there were errors).     That way you'd never have the issue you noted.

 

Well, it is March 1st, and my monthly parity check will run tomorrow night...so I'll leave it set to correcting and see what happens. It'll take 16 hours, but I'll post after it completes. Thanks much for the responses.

 

It has always said zero errors until my stupid fingers got involved...curse us and splash us.

Jeez - I promised an update and FORGOT.

 

Everything was good after the next parity check - back to zero and all is well! Now I need to decide if I wanna go to NON correcting checks only. Before I go buttheaded again.

Thanks for everything!