Just moved to different hardware (TAM solutions H8DME-2 board with opteron CPUs) after testing the new hardware with a simple unraid setup and a few spare drives.  Worked fine so moved my old parity and data disks to the new system. All was working fine on 5.05.  Passed a parity check and accepted new files. Memtest was good. MOVER was fine, etc.  Then I had some sync errors on a parity check (3100 !, never had 1 on old hardware) so I did a memtest (good) and powered down to check cabling and swapped out a SATA card that I had suspicions about.

 

I must have bumped something as one drive came up redballed (could have been orange) when I booted back up.  I could not get unraid to accept the drive. I checked the cables, same redball, but I know unraid seems to remember a drive state if it saw a problem real or imagined.  So I put in a new drive that showed as blue balled so I think the connections are OK.  I put the drive in a diff slot in a separate drive cage, same issues.

 

I put the original drive back in.  On reboot the old drive was still redballed.  I did get it to show greenball at one point - can't remember the magic that made that happen, but it wanted to check parity and it was incredibly slow (Kbs!) so I stopped.

 

I rebooted again hoping it would just all be fine and unfortunately the redball (maybe orange) was back.  My parity drive also shows an orange ball.  I'm on 5.05 so I tried the new config button and reassigned the drives.  When I started the array, the problem drive now shows as unformatted and the parity is invalid.  After plowing through the forum I started in maintenance mode and ran rieserfsck on the "unformatted" drive.  It says there's a hardware problem on the drive in that it "cannot read the block (2) input/output error".  It passes SMART. It also printed a paragraph about bad sectors and says to replace the drive.  Never ran this before so I don't really know what to make of it.

 

I put in a new replacement drive thinking that I could rebuild the data and if things went bad I would still have the original drive for data recovery.  The array sees the drive and asks to format it.  I initiate formatting and nothing happens, meaning that if I hit refresh after awhile it goes back to the screen showing the drive as unformatted.  Plus there is no disk activity.

 

For now I powered down the array.

 

I'm a bit of a noob (unraid has been so reliable that I have not had to gain much experience).  I can supply syslogs, but I need to know what configuration (old drive, new drive) would be useful and the things to try (format) etc so the right info is in the log.

 

 

 

update:  Installed YAReg on a PC and all data appears to be there on original "failed" data drive.

If you put in a new drive and the parity is invalid it won't replicate the faulty drive. It sounds like somehow you parity was invalid and a data drive failed at the same time. Worse case is you have lost the data on the failed drive. Maybe a more experienced user will chime in but it sounds like you are going to have to do a new config, assign all the drives and let it build new parity data. Did you run a preclear on the new drive before adding it to the array?

Make sure the new disk is smaller in size than your parity disk. If you just added a new drive that wasn't pre cleared then I can imagine it will take a long time before the webGui responds after selecting format disk, see this:

http://lime-technology.com/wiki/index.php/UnRAID_Manual_5#Add_one_or_more_new_disks

 

Thanks for the reply

 

I did not preclear the new drive.  It did not give me the option and I have not got the preclear script running.  The data appears to be on the old "failed" drive as I can see and access it on a PC when I hook up the drive and use YAReg.

 

I'f I could get the array back up with good parity (even without the old failed drive) one option might be to copy the data from the old drive to the unraid arrray via my my PC.  I'm not sure that would work.

 

I think I will wait for more advice as you recommend.

You should always preclear a drive before adding it to your array to check its condition. I run one many others run more. 30 hrs on a 3TB. You should read up about preclearing drives in the forum. If you just go ahead and add and then format the drive without testing it then you could be asking for trouble. For now I would leave the new drive out of the array and start it up. Do a new config and assign your old data drives and then the parity. Start the array and let the parity build and then once you have it running add a new precleared drive to the array and transfer your old data across the network if you can. Setup a monthly (Some do it more often) parity check in the WebGui and monitor any errors showing up in the webGui after it has done its monthly check.

Thanks. I've always precleared new drives in the past, but that was in part due to unraid doing it automatically.  I've always done regular parity checks - since 2006!

 

So here is what I understand I should do:

 

Start unraid

stop the array

click new config

assigned parity and old data drives - question here: do the drives need to have their original drive numbers?

do a parity check

 

then:

 

Add new drive (format, preclear)

copy data back from "failed" drive to new drive over the network

 

I'm OK with this, but it seems there should be a way to get unraid to read a drive that YAReg can read and just pull it back into the array.

This question depends:

"assigned parity and old data drives - question here: do the drives need to have their original drive numbers?"

This depends on what version you are running. Not if its V5.05 like you said and it shouldn't matter anyhow if your rebuilding a new parity, but make sure you select your old parity drive and not a data drive. I presume your sync errors you spoke about were from your failed data drive and not your parity drive?

After you assign your drives in the new config. Start the array and it will rebuild your parity based on the current data drives.

I also didn't know unraid ever precleared drives automatically. I though Joe's script was always run from the command line.

My one fear are those parity errors as I was diagnosing their cause when the "failed" drive when bad.

 

I'm moving all the data off my "failed" drive to my PC for now to be sure the data is recovered before doing anything on my unraid array.  This will take overnight I suspect.

 

I'll report back but am open to more suggestions.

 

Thanks for the help.

 

 

If you are worried about the possibility of a failing parity drive as well why not replace it with your new drive and then rebuild the array. Any files on the data drives is safe as long as the data drive is not faulty. Then preclear your old parity drive to see if you get re-allocated sectors or other errors. Of course this all depends on your parity and data drive sizes, remember the parity drive always needs to be larger or at least the same size as your largest data drive.

Mostly worried about a sata card not a drive. 

 

Am I correct that the parity check process is a one-way street so to speak in that the data on the data drives stays intact and adjustments are made on the parity drive.  So if the parity drive fails the data is OK, but vulnerable to failure of one of th edata drives.

 

 

 

Data is read from the data drives and written to the parity drive.

Just to draw closure on this.

 

I followed my intended procedure after pre-clearing a new 2TB drive external from the array and then added it following:

 

Start unraid

stop the array

click new config

assigned parity and old data drives

do a parity check

add new drive

 

All went according to plan except that once I added the new drive the array began to rebuild the drive.  It wrote to the drive for over 10 hours only to come up with no data on the drive.  I have no idea what all the writing was about.

 

The array now seems fine and I will copy the data I recovered over from my PC

 

 

When you added the drive and done a new config and started the array it would have read all the data drives and wrote to the parity drive to protect the other data drives in the array and the new drive you added even though it was empty. By doing a new config you basically tell the parity drive that it is invalid and thus when you start the array the parity drive has to build itself from the data drives in the array and once it was finished all your data drives are safe from a single data disk failure.

Thank you for all of your help megalodon, but I was fine with the parity rebuild.  What was odd was the data drive rebuild after the parity reconstruction when a added a new (precleared -) data drive.

 

I may have spoke too soon about things being SOLVED.  My shares have not appeared as folders on my new data drive.  I checked and I do not have any drives listed in the include or exclude under settings/shares/.  Is there something new in v5.05 to get the shares onto the new drive?  It used to be automatic.

It was probably just reading the data drive not rebuilding it. If you done a new config if won't rebuild anything just read all the data drives and write to parity. Your shares depends on you file allocation method, I use high water and if I drag a folder called rambo to my main movies share and say disk 4 was new and was about to be used it would create a folder called movies and place rambo inside of that folder for me. You don't have to create them manually unless you drag folders to the individual disks.

Ok so a new drive wil not show shares until first used?  I too use high water.

Correct.