Hello,

 

I've been running my unraid server for 6-7 years and have never had a problem with it.  I recently started filling up my data disks so I upgraded my parity drive and one data drive.  The latter which resulted in my parity drive becoming disabled.  Before I do anything stupid, I'd very much appreciate a heading check on this.  My original configuration was: 

 

2TB parity, 2 x 1TB data disks, 3 x 2TB data disks (all green balls)

 

I bought 2 x 4TB drives and precleared them both using the preclear plugin and the default settings and both drives passed.  So I swapped my 2TB parity disk for a 4TB disk and rebuilt and then rechecked parity.  Everything was great.  So my intermediate configuration was:

 

4TB parity, 2 x 1TB data disks, 3 x 2TB data disks (all green balls)

 

Then I removed one of the 1TB data drives and replaced it with the other 4TB I had just precleared.  Unraid noticed I was missing the 1TB drive, so I assigned the 4TB drive to that disk and started the array.  It began a prarity-sync/ data rebuild.  Everything seemed to be going fine until ~50% of the way through the rebuild when my parity drive started racking up errors like crazy.  The parity sync ended with 495M errors.  Attached are the notifications I got at the end as well as my syslog and the SMART reports.

 

The server has been sitting on my desk plugged into the same power supply it has been for the last couple days and the power cables and sata cables are snug in the drives and the MB.

 

Current configuration is:

 

4TB parity, 1 x 1TB data disks, 3 x 2TB data disks and 1 x 4TB  (all data disks are green balls, but the parity disk is a red X and disabled)

 

Did I do something wrong when I swapped out the 1TB data disk with my new 4TB drive?

Is it a coincidence that the rebuild started producing error when it got passed 2TB (the former max size of my drives)?

Is it normal to get errors like that when expanding a data disk beyond the former max size of the other data drives?

Should I recheck/rewirte parity now that the new disk is showing green?

I haven't done anything yet, and don't plan to until I can confirm that is a good idea and won't lead to irreversible data loss.

 

Note:  I still have my original 2TB parity drive and the old 1TB data drive untouched since I removed them.  So I can go back to my original configuration if needed I suppose.

 

Many thanks for reading this and for any help you might be able to provide!

 

 

syslog.txt

ST4000DM004-2CV104_ZFN1MRTV-20190206-2026 parity (sde).txt

WDC_WD10EARS-00Y5B1_WD-WCAV5F079658-20190206-2026 disk3 (sdf).txt

WDC_WD20EARX-008FB0_WD-WCAZAE351759-20190206-2026 disk1 (sdd).txt

WDC_WD20EARX-008FB0_WD-WCAZAH409009-20190206-2026 disk2 (sdc).txt

ST2000DM006-2DM164_Z4ZCJSD3-20190206-2026 disk5 (sdb).txt

ST4000DM004-2CV104_ZFN1M5V8-20190206-2026 disk4 (sdg).txt

I haven't looked at any of those attachments. You can get me all of that and more in one simple attachment, much easier for you to do, and much easier for us to work with. Please always do it this way:

 

Go to Tools - Diagnostics and attach the complete diagnostics zip file to your next post.

Sorry about that, figured that might be easier.  The .zip from the diagnostics is attached.

tower-diagnostics-20190206-2026.zip

Looks like a bad connection to parity. It isn't even reporting SMART. You must always double check all connections whenever you are mucking about in the case.

 

Looking at syslog, I think you are probably right that parity disconnected after the rebuild was beyond the size of the original disk, and the Notifications screenshot seems to agree, so you are probably OK to just rebuild parity. Hang on to that original data disk just in case.

 

Some other things I noticed. Your syslog is being filled up with some useless messages. This makes it much harder to read.

 

Feb  6 09:55:03 Tower emhttpd: error: get_key_info, 581: No such file or directory (2): get_message: /boot/config/._Plus.key (-3)

^This one comes from having an invalid .key file. Apple ecosystems have an annoying habit of dropping hidden "._" files on every thing they touch. Delete the file "config/._Plus.key" from your flash drive. This one isn't too bad since it only tries to read the key when you start the array.

 

Feb  6 10:01:59 Tower root: error: /plugins/preclear.disk/Preclear.php: wrong csrf_token

^This one is caused by leaving an old browser session open after rebooting. It might be on another tab, another computer, or possibly on a mobile app. You should close all those. This one is especially annoying since it just keeps going and going for page after page.

 

 

And you are running an old version of Unraid. You should upgrade. For one reason, newer versions give us even better diagnostics to work with.

Thanks.  I do have 2 browser tabs open the unraid GUI on 2 separate machines.  One machine is new and I haven't finished switching everything over yet.  I checked the cables after I got the errors and everything is tight.  Once I get this unraid upgrade figured out I will delete that file from the flash drive.  My old machine is an iMac so that is definitely the culprit with the hidden dot files.

 

I am running unraid version 6.4.0.  I don't mess with upgrading it too often because it's been performing, rock solid, for so long.  I'll look into updating.

 

One additional note as I've been re-reading through the wiki; when I swapped in the last 4TB data drive and started the array, I can't recall specifically checking the "Yes I want to do this" box when I started the array (I don't recall seeing it, or if it was there it might have already been checked since it started the parity-sync/rebuild?). 

 

If that box was not checked would that result in the parity sync errors I got?  

 

To be clear, you recommend I rebuilt the parity now,  1 -Shutdown the server, 2- Restart the server, 3 - start the array, and 4 - rebuild parity (write corrections checked)?

6 minutes ago, dmgibney said:

I can't recall specifically checking the "Yes I want to do this" box

There isn't a box to check when replacing a drive.

 

6 minutes ago, dmgibney said:

you recommend I rebuilt the parity now

Parity is disabled currently, so you will have to make Unraid think you have replaced it.

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

 

Got it.  I'll do that now and let it rebuild over night.  I will report back in the morning with all green balls (hopefully).

 

Thank you for your help!

Looking at those Notifications again the rebuild wasn't complete until after the errors started. But if you are sure it was already beyond the size of the original it should be OK.

A few obs:

Parity needs a new SATA cable

Onboard SATA controller is set to IDE, that board might not have AHCI support, but check the bios, it it does change to AHCI.

Rebuild wasn't finished when the parity errors started but it was past the data part of the rebuild, so rebuilt disk should be fine.

Ok, the parity-sync finished and reported 0 errors this time.  I used the same SATA cable since I don't have a spare one (yet).  The array is showing green balls on the data drives and now also on the parity drive.  (syslog.zip attached if curious)

 

It was a bit strange while following the steps from above to unassign and then reassign parity.  When I went to reassign it, my new 4TB drive showed up for an instant then disappeared.  I started and stopped the array again and went to reassign and it didn't show up at all.  So I rebooted the server and it showed up like normal.  I didn't touch any cables at all...

 

I have a this MB (https://www.newegg.com/Product/Product.aspx?Item=N82E16813182162).  Looking at the documentation it doesn't mention AHCI so I'm guessing it doesn't support that.  What is the advantage of AHCI?

 

You both have stated that the rebuilt data drive 'should be fine' since the parity errors started after it was passed the 1TB point of the original data drive.  Aside from just checking for random lost/corrupt files is there any way to be sure it is OK?

 

Again, thanks your help getting this figured out!

tower-diagnostics-20190207-1106.zip

16 minutes ago, dmgibney said:

What is the advantage of AHCI?

There are some performance advantages, but mostly in IDE mode when there are issues on one disk it can also take out the respective master/slave.

 

17 minutes ago, dmgibney said:

Aside from just checking for random lost/corrupt files is there any way to be sure it is OK?

Not unless you have checksums of your files or were using btrfs, though if you still have the old disk you could run a file compare utility.