Hello,

 

I was in process of doing a parity swap when one of the drives came up with an error (Current pending sector). I took that as the drive is about to fail. I also noticed that during the parity rebuild that it stopped reading from that drive. I canceled the parity rebuild as I figured it would no longer be useful.

I tried to select the old parity drive as my parity but it says parity is invalid. How do I go about in putting back my old parity drive?

 

Thanks,

Ayman

Please post the diagnostics.

Attached 

plexpogo-diagnostics-20230505-1046.zip

Diags are after rebooting, but parity is invalid and there's no disabled disk, that doesn't track with the parity swap procedure, maybe you were doing a parity upgrade?

I must be using the terminology wrong. I wanted to upgrade my parity disk size, so I disconnected my parity drive and then swapped the drive to the larger one. I then let it rebuild the parity. That is when during the parity rebuild, I got a notification saying one of the disk drives (disk 7) had a current pending sector. I then stopped the parity rebuild. 

I assumed the next step would to reconnect the old parity drive (this is where I am getting hung up on), so I then can replace the failing disk drive. Is that the wrong thought process?

Edited May 5, 20233 yr by zalamah

You can do a new config and trust parity but this will only work if nothing was written to the array since old parity was removed, even so, and assuming arrays was online old parity will no longer be 100% in sync, but it's usually fixable, if the array disks are xfs, to try:

 

-Tools -> New Config -> Retain current configuration: All -> Apply

-Check "parity is already valid" and start the array

-Stop array, unassign the disk you want to replace

-Start array, post new diags.

One pending sector is an early warning, shouldn't be an issue. 

 

52 minutes ago, zalamah said:

I also noticed that during the parity rebuild that it stopped reading from that drive.

That's probably because it's a smaller drive than others, once parity has gone past the 6TB then of course it has nothing to read from it anymore. Where was it when you stopped?

 

Old parity would only be valid if everything since the change was done in maintenance mode so nothing got written to the array. If it wasn't I'd just restart the parity build on the new drive, probably shouldn't have been stopped in the first place.

Edited May 5, 20233 yr by Kilrah

Before I do this, I have not yet received my replacement hard drive for the failing drive. I am hoping it will come tomorrow. Should I wait until then before I do anything or proceed with everything anyway since it doesnt change the outcome. 

6 minutes ago, Kilrah said:

One pending sector is an early warning, shouldn't be an issue. 

 

That's probably because it's a smaller drive than others, once parity has gone past the 6TB then of course it has nothing to read from it anymore. Where was it when you stopped?

 

Old parity would only be valid if everything since the change was done in maintenance mode so nothing got written to the array. If it wasn't I'd just restart the parity build on the new drive, probably shouldn't have been stopped in the first place.

 Ahhh, you are right! It was after it the 6TB. Ugh. i did not put two to two together

Then my next question, should I just let it rebuild the parity with the larger drive again? Once that is done, I can then replace the "bad" drive if I wanted to?

I would go for it, as you say it won't change the outcome. And also probably not rush replacing that drive unless smart gets worse. That pending sector may even disappear on a subsequent write to that location. 

1 minute ago, Kilrah said:

I would go for it, as you say it won't change the outcome. And also probably not rush replacing that drive unless smart gets worse. That pending sector may even disappear on a subsequent write to that location. 

Just to confirm go with the new config or just let it rebuild? I assume the latter is the safer bet?

You can run an extended SMART test to confirm, but unless it's a "false positive" disk will have read errors during a parity sync.

My current plan is to run an extended test of disk7. Found out the results of that.

If no errors, then i will just rebuild the new parity drive. 
If there are errors, put back the old parity drive (hopefully this works), replace disk7 with a new hard drive and rebuild that. Then do a parity upgrade and then use the old parity drive as another disk to the pool. 

12 hours ago, JorgeB said:

You can do a new config and trust parity but this will only work if nothing was written to the array since old parity was removed, even so, and assuming arrays was online old parity will no longer be 100% in sync, but it's usually fixable, if the array disks are xfs, to try:

 

-Tools -> New Config -> Retain current configuration: All -> Apply

-Check "parity is already valid" and start the array

-Stop array, unassign the disk you want to replace

-Start array, post new diags.

 I followed these directions and attached the new diag. My replacement drive for disk7 is coming in tomorrow. 

 

I assume next steps would be to put new drive for disk 7 and let it rebuild. once that is done, I can do a parity upgrade. Then I can add the old parity drive as another drive.

 

Is this correct? 

plexpogo-diagnostics-20230505-2337.zip

So disk7 is confirmed bad and you already did the new config with the old parity, did you start the array to confirm emulated disk7 is mounting? That's not in the diags.

I started the array in maintenance mode initially with disk7 assigned but then I stopped the array and unassigned it.

 

Should I reassign disk7, start the array and mount all the disks? Did you want the diagnostics at this point? 

 

Wasn't sure if you wanted me to start the array normally with All the disks. My replacement drive is currently out for delivery.

 

If disk7 mounts just fine, should I then use unbalance plug-in to move all the data from that drive to the replacement drive? Then in theory, create a new parity drive (ie I can use my 18tb drive) ? 

 

Or should I just let the parity drive rebuild "disk7"?

 

You should start the array with disk7 unassigned to see if the emulated disk is mounting.

Does not seem like it's being emulated.

 

I had to do a new config to start array with the disks mounted. Then I stopped the array and unassigned disk7. Started the array again with disks mounted.

 

Attached are the diagnostics. 

plexpogo-diagnostics-20230507-0419.zip

Check filesystem on the emulated disk 7.

Kind of confused. It is not being emulated though. But it is xfs.

 

Do you want me to click on "check" under the check file system status? 

4 minutes ago, zalamah said:

Kind of confused. It is not being emulated though. But it is xfs.

It MUST be emulated as there is no physical device assigned to disk7.

Therefore run the check filesystem on disk7 as rebuilds do not fix file system level corruption so we want to know if the emulated drive can be repaired.

12 minutes ago, zalamah said:

Do you want me to click on "check" under the check file system status? 

Yes, and remove -n

I clicked on check several times and nothing is happening as far as I can tell.

I am beginning to think I must have messed something up.

I feel like just mounting the bad drive and offload the data onto another drive and just create a new parity drive would be the best direction going forward. How does one create a new parity drive? Just assign a new drive to the parity slot and let unraid build it? 

Start the array in maintenance mode and post the output of:

xfs_repair -v /dev/md7