Disk disabled on Unraid 6.9.2

[solid-eml]

One of my disks are disabled and I need help to fix it.

A year ago I had UDMA CRC errors which I concluded where caused by a bent SFF-8087 cable, which I thought I fixed. 

I couple of weeks ago I got the same error, and after a couple of days the drive was disabled. Now I have changed the SFF-8087 cable with a new cable, but the drive is still disabled. The SMART test I passed, and I don't understand the results of the xfs_repair status.

I am not sure how to proceed. Any help on how to fix the issue?

silverwolf-diagnostics-20210425-1558.zip xfs_repair status.txt

Edited April 25, 2021 by solid-eml
Spelling error

[itimpi]

The only way to clear the ‘Disabled’’ staus is to rebuild the drive as described here in the online documentation available via the Manual link at the bottom of the unRaid GUI.

 

If you think the drive is fine you can use the option to rebuild onto itself.

 

Was there a specific reason you ran the xfs_repair?    It did not indicate any obvious problem, but since the ‘no modify’ flag was set it would have run in read-only mode and made no changes.   The commonest reason would be because a drive suddenly showed as ‘unmountable’ indicating it was likely some file system corruption had occurred.

[solid-eml]

Thanks for the reply! I was hoping maybe there was another solution then rebuilding, but if not I'll go ahead and do that. 

 

I ran the xfs_repair was because read it might be a possible solution with unmountable disks. Perhaps I misunderstood.

 

Anyway, I'll try do rebuild and see where that takes me. Thank you for helping!

 

[itimpi]

8 minutes ago, solid-eml said:

an the xfs_repair was because read it might be a possible solution with unmountable disks. Perhaps I misunderst

That is the correct thing to a disk showing as ‘unmountable’.   However you said the disk was showing as ‘disabled’ - not as unmountable?    They are different independent conditions.

[trurl]

6 minutes ago, itimpi said:

That is the correct thing to a disk showing as ‘unmountable’.   However you said the disk was showing as ‘disabled’ - not as unmountable?    They are different independent conditions.

And since your diagnostics were taken with the array stopped, we can't tell whether your disks mount.

 

Start the array and post new diagnostics if you want any advice about that.

[solid-eml]

The disk is marked as disabled. I attached a screenshot showing the error. I also attached the diagnostics with the array started.

silverwolf-diagnostics-20210425-2022.zip

[trurl]

Screenshot doesn't show whether the emulated disk2 is mounted though. That information would appear to the far right which you have clipped out. It would show "unmountable" with no size information in any columns if it were not mountable.

 

And your diagnostics do indeed show that emulated disk2 is mounted. You should be able to access its files as if the disk were actually being accessed, but it would really get the data from the parity calculation by reading all other disks.

 

SMART for disk2 looks OK, but it also shows you haven't completed an extended SMART test. I think there has been an issue with the disk spinning down when trying to do the test, so you may have to set the disk to not spin down while the test runs.

 

Since you rebooted before getting any diagnostics we can't see why the disk became disabled but a connection issue seems likely.

 

You can try the extended SMART again as mentioned if you want. It will take many hours with 8TB.

[solid-eml]

58 minutes ago, trurl said:

Screenshot doesn't show whether the emulated disk2 is mounted though. That information would appear to the far right which you have clipped out. It would show "unmountable" with no size information in any columns if it were not mountable.

 

And your diagnostics do indeed show that emulated disk2 is mounted. You should be able to access its files as if the disk were actually being accessed, but it would really get the data from the parity calculation by reading all other disks.

 

SMART for disk2 looks OK, but it also shows you haven't completed an extended SMART test. I think there has been an issue with the disk spinning down when trying to do the test, so you may have to set the disk to not spin down while the test runs.

 

Since you rebooted before getting any diagnostics we can't see why the disk became disabled but a connection issue seems likely.

 

You can try the extended SMART again as mentioned if you want. It will take many hours with 8TB.

 

In disk settings the disk I label as mounted.

 

I'll disable the spin down delay and to an extended SMART test overnight. If it passes I'll try to remove and add the disk back to the array to rebuild.

Should I delete the partition before adding it to the array? And should I do a preclear to see if there is any read/write errors?

 

 

[trurl]

2 hours ago, solid-eml said:

Should I delete the partition before adding it to the array? And should I do a preclear to see if there is any read/write errors?

No point. Rebuild will overwrite the entire disk.

 

I don't like the word ADD in this situation, since you are not ADDING a disk, you are REPLACING a disk with the same disk. Makes me wonder if you know how to proceed. Do you know how to rebuild to the same disk?

[solid-eml]

6 hours ago, trurl said:

I don't like the word ADD in this situation, since you are not ADDING a disk, you are REPLACING a disk with the same disk. Makes me wonder if you know how to proceed. Do you know how to rebuild to the same disk?

I am sorry for using the wrong terminology. But yes, I belive I do know how to rebuild to the same disk:

Stop the array.

Remove the disk for the array.

Start the array so it will be registered as missing.

Stop the array.

Reassign the disk.

Start the array to begin rebuilding.

Wait for the rebuild to complete.

[trurl]

7 hours ago, solid-eml said:

Remove the disk for the array.

And here I would say "unassign the disk". Some people might interpret "remove" as meaning the disk actually had to be taken out.

 

Sounds like you are good to go.

[solid-eml]

2 minutes ago, trurl said:

And here I would say "unassign the disk". Some people might interpret "remove" as meaning the disk actually had to be taken out.

I understand someone might misinterpret the different terms. I'll try to use a more precise language in the future.

Anyway, thanks for helping out. I am currently four hours into the data-rebuild, and it will be some time until it is finished. But it looks like it solved the problem.