Parity Check Errors

[ZeroThirtySeven]

This is a New Build.  After Getting everything setup the way I wanted, I started to put Data on my server.  I then ran a Parity Check.  It says that Parity is ok, but that there are 700 MIllion Errors.



Is this an Issue with the disk?  The dashboard says it's healthy

Will I have an Issue with getting my Data if I were to take a disk offline or Replace a disk?

diagnostics-20210831-1621.zip

[trurl]

Did you ever let parity build when you first assigned parity? Did you ever do New Config and not let it rebuild parity?

[ZeroThirtySeven]

5 minutes ago, trurl said:

Did you ever let parity build when you first assigned parity? Did you ever do New Config and not let it rebuild parity?

Yes to Both.

 

I had some issues with getting 1 of the disks to format and Mount.  So I used the "New Config" several times in my Trouble shooting.  After I got all 6 Mounted, a Parity check finished, but I don't recall if there were errors.

 

[trurl]

1 minute ago, ZeroThirtySeven said:

So I used the "New Config" several times in my Trouble shooting.  After I got all 6 Mounted, a Parity check finished

Your answer was not a clear and unambiguous "yes" to the New Config part of the question.

 

When you New Config, there is a check box you can check before starting the array, which tells Unraid parity is already valid, and so it doesn't rebuild parity after the New Config.

 

You said "parity check finished", but a parity check isn't the same thing as a parity rebuild, which is what I was asking about.

 

[ZeroThirtySeven]

Parity Rebuild was not done has I had not put any data on the drives.  This is was a new setup using the trial version of the software. 

 

Several times I would use the New Config when ever UnRaid failed to Format and mount the drive in disk 1.  I found out the issue was, is that the previous owner of the hardware had several of the SATA ports set to IDE emulation.  Once I made that change, I used the New Config to start over again, and then I was finally able to format and mount the disk in Drive 1.  I then continued to setup the Server by creating shares and then using thoes shares.

Did I hit that Check mark?  I don't recall.

Edited August 31, 2021 by ZeroThirtySeven

[JonathanM]

2 minutes ago, ZeroThirtySeven said:

Parity Rebuild was not done has I had not put any data on the drives. 

Parity doesn't work with data, it works with raw bits, which are typically a jumbled mess on used drives that haven't had a lengthy process done to set every bit to a known value.

[trurl]

I guess it doesn't matter. The only sure thing to do is rebuild parity, then check it again after. If there are any hardware problems rebuilding parity will make those apparent.

1. Stop array
2. Unassign parity disk
3. Start array with parity disk unassigned
4. Stop array
5. Reassign parity disk
6. Start array to begin parity rebuild

During parity rebuild there should be a lot of reads of all data disks, and a lot of writes to parity, and zeros in the Errors column. Post new diagnostics if there are problems during parity rebuild, or post new diagnostics after rebuild completes. Do not reboot because we need to see syslog for the rebuild and for the subsequent parity check.

[ZeroThirtySeven]

Yea the Drives were definitly used before I got them, so they definitly were not zero'ed out.

 

6 minutes ago, trurl said:

1. Stop array
2. Unassign parity disk
3. Start array with parity disk unassigned
4. Stop array
5. Reassign parity disk
6. Start array to begin parity rebuild

Is there any risk of Data loss performing these Steps?  And Should I wait until I finish moving my data over or stop moving Data and do these now?

[trurl]

Parity contains none of your data and no disk except parity is written by parity rebuild, so no risk to rebuilding parity except for the fact that you don't currently have valid parity so if you do have a problem with a data disk you might not be able to recover it.

 

Many people do the initial data load without parity since parity writes slows things down. If you have the original source data then not much risk to running without parity for the initial load.

 

If you want to run without parity, just do steps 1-3, then when finished with the initial load, do steps 4-6.

[ZeroThirtySeven]

Thanks For the Help!  I will do these ASAP.

Just so I am learning something, the Party Errors that i am seeing is because the first Parity check happened when I had random 1's and 0's on the drives?

[trurl]

The only way parity might be valid with new disks is if you had precleared all the disks including parity. Then they would all have nothing but zeros on them, and 0+0=0

[ZeroThirtySeven]

Is this Correct?  No Device the same as Unassign?

[trurl]

1 minute ago, ZeroThirtySeven said:

Is this Correct?  No Device the same as Unassign?

yes

[ZeroThirtySeven]

Ok! Thanks for the Help, this looks like it will take awhile.  I will let you know!

[trurl]

That screenshot is pretty soon after starting so might not be a good indicator of what the speed will be.

 

I usually estimate 2-3 hours per TB of parity unless you have bottlenecks like port multipliers, so it should complete a lot sooner than it shows there.

 

Just noticed a Marvell controller in your diagnostics. Those are not recommended but leave it for now and see how it goes.

 

Is the Errors column on Main all zeros?

[ZeroThirtySeven]

In My Previous issue with the DISK 1 slot not mounting/Formatting, the Marvel Controller was already brought up and I found that that Motherboard only uses thoes for the eSATA which are not being used in my configuration.  I believe it is the standard AMD controller for the Motherboard SATA.  

The Speed definitely picked up after I posted.  I have a bout 4 hours remaining.  I will most likely not be awake when it finishes.

 

I do not see it listing any errors as part of the Parity-check/Data-rebuild.  There are also currently no disk errors listed.

Edited September 1, 2021 by ZeroThirtySeven

[trurl]

That's more writes than I would expect to the data drives. Nothing necessarily wrong with that. Are you still loading data while building parity? Nothing wrong with that either but it would slow things down.

[ZeroThirtySeven]

I didn't Clear the stats.

 

I also lost power after I had gone to sleep.  Upon booting back up, it started doing another Parity-sync/Data-Rebuild.  So I am starting over.

[JonathanM]

1 hour ago, ZeroThirtySeven said:

 

I also lost power after I had gone to sleep. 

UPS?

[trurl]

1 hour ago, ZeroThirtySeven said:

Upon booting back up, it started doing another Parity-sync/Data-Rebuild.

Does it actually say that? Just want to make sure it isn't only doing a non-correcting parity check due to unclean shutdown.

[ZeroThirtySeven]

 

I have gotten Errors on Disk 5 however.

[trurl]

46 minutes ago, ZeroThirtySeven said:

I have gotten Errors on Disk 5 however.

Unless that gets worse let it continue. We will do a parity check after anyway.

 

Get us new diagnostics after rebuild completes (or if it looks like it is going bad) and don't reboot.

[ZeroThirtySeven]

Ok, it Finished with 57 Errors

 

diagnostics-20210901-2234.zip

[trurl]

Disk5 seems to have gotten a pending sector during that.

 

Disable spindown on disk5 and run an extended SMART test on it. It will take several hours.

[trurl]

Just noticed that all your WD disks have nonzero SMART attribute 1 (Raw Read Error Rate), which isn't monitored by default but should be added to the monitored attributes for WD.

 

Disable spindown on all those disks and run extended SMART tests on them.