Parity check errors?

[Dave001]

Hey,

      Using Unraid v5.0.4, it done the auto parity check this morning, and came up with 2151 errors, I don't recall ever seeing a single error before, so I'm guessing somethings not right.

 

Any help would be appreciated.

 

Thanks,

Dave

 

syslog.zip

SMART_Tests.zip

[garycase]

Assuming you ran a correcting check, run it again and confirm there are now zero errors.

 

[Dave001]

I have the monthly check set to "NOCORRECT" , nothing has been corrected yet.

 

Should I just go ahead and run a correcting parity check?

 

 

 

 

 

 

 

 

[garycase]

I would.

 

If you want to confirm the errors are indeed on the parity disk (most likely) you could first do a full comparison of all of your data against your backups.

 

You could also check that all your cables are firmly seated and any disks in hot-swap cages are firmly in place ... loose connections can cause errors.

 

Are there any read errors indicated in the report?

 

In any event, I never bother with non-correcting checks.    If there's an error, I want it fixed.    The first few times I had sync errors I ran a complete comparison of all my data afterwards to confirm the errors had indeed been just sync errors -- and in every case that's always been true.

Been a long time since I've had any sync errors, but if I had them now, I could test things a little easier, as I now maintain checksums of all the files in addition to my backups ... so I'd just verify the checksums if I had errors now.  But as I noted, ALL of my parity checks are correcting.

 

 

[Dave001]

Run the correcting parity check, which fixed the 2151 errors, then run another check, no errors found.

 

Will see how it goes from here.

 

Thanks for the help.

 

Dave

 

 

[garycase]

You're welcome.

 

This topic comes up enough that I decided to write a long post detailing my thoughts on it, which you might want to read.

 

Since what you can best do when you have these errors is related to whether or not you have good backups, I've added it to a similar post I wrote on backups.  You can read them both here (the first comment is on backups; the second on parity checks):

http://lime-technology.com/forum/index.php?topic=31020.0

 

[Dave001]

While I'm sure most, like me, appreciate the time you took to write that explanation, you're failing to take one thing into consideration, sometimes not everything stored on the server requires backing up.

 

I have around 9TB of data on my server, but only about 2TB is important to me and backed up. If the rest was lost, it is not a big concern. The time it would take to compare that 9TB of data with the backups, even if I did have them, would not be worth it to me. I care about it enough to devote an extra drive as parity, in the chance that if a drive dies, I should be able to recover the data, but if It can't, it's not a big deal.

 

My post was more asking where the errors could have come from, the smart reports don't list any faults, all the Sata cables are quality cables that clip into the drives/motherboard headers, the PSU is more then capable of supplying the required power, and all power outputs are well within spec. I have not had an unclean power down in the last couple of months, so I was unsure where to look for the fault.

 

While I agree that you are crazy not to have any important data backed up, and relying on the parity disk as your only backup is a bad idea, I also think it's just as crazy to preach that everyone should keep a 100% backup up of all the data on the server. Backup your important data, offsite preferably, but why waste storage space and time backing up unimportant data?

[garycase]

Perhaps you missed #9 in my write-up:  "(9)  Nobody NEEDS backups.  ... But if it would upset you to lose it, it should be backed up  "

 

If you keep data on your server you don't really care about, then clearly it doesn't need to be backed up.    But if you still want the ability to know if it's been corrupted, you should at least maintain checksums of those files, so you'll have a way of knowing what's good and what's not.

 

The key point I was trying to make in my write-up is that UnRAID -- or ANY RAID system -- is NOT a backup.  If you care about your data, back it up.    If you don't, then I agree with you that you don't need to back it up ... for that matter, you probably don't need it on a fault-tolerant array.