Possible to remove an new istalled datadisk to replace it with an failing disk ?

[Nortrek]

Hi

 

I have an failing disk, and i installed an fresh new disk yesterday that have not been filled with data yet.

 

How woul`d i go forward to remove the new disk i installed, and then remove the failing disk and install the new disk in this slot ?

 

EDIT:

Can i unmount the failing disk and the new empty datadisk, then insert the new empty datadisk instead of the failing one and do an new config and let the parity rebuild the raid again ?

 

Regards

Nortrek

[itimpi]

Hi

 

I have an failing disk, and i installed an fresh new disk yesterday that have not been filled with data yet.

 

How woul`d i go forward to remove the new disk i installed, and then remove the failing disk and install the new disk in this slot ?

 

EDIT:

Can i unmount the failing disk and the new empty datadisk, then insert the new empty datadisk instead of the failing one and do an new config and let the parity rebuild the raid again ?

You cannot do a New Config as that would discard parity and thus the failing disk would be unrecoverable.  I do not think that there is any safe way of doing what you have asked for as there is no easy way to reduce the number of disks without invalidating parity.  The only safe way is to get a replacement for the failing disk.

 

One question - what makes you think that the drive is failing?  It is always possible that there was a temporary glitch on the drive rather than it actually being a failing drive?  It might be worth getting a diagnostics report (via the Tools tab)  and posting that here as that would show the SMART reports for the drives, and also the syslog for any error messages.

[Nortrek]

Hi

 

Here is the photo i posted from my other post:

 

 

I also attatched the Diagonostic report as an attatchment to this post.

 

Thank you very much for the help

tower-diagnostics-20150703-1526.zip

[SSD]

Hi

 

I have an failing disk, and i installed an fresh new disk yesterday that have not been filled with data yet.

 

How woul`d i go forward to remove the new disk i installed, and then remove the failing disk and install the new disk in this slot ?

 

EDIT:

Can i unmount the failing disk and the new empty datadisk, then insert the new empty datadisk instead of the failing one and do an new config and let the parity rebuild the raid again ?

You cannot do a New Config as that would discard parity and thus the failing disk would be unrecoverable.  I do not think that there is any safe way of doing what you have asked for as there is no easy way to reduce the number of disks without invalidating parity.  The only safe way is to get a replacement for the failing disk.

 

One question - what makes you think that the drive is failing?  It is always possible that there was a temporary glitch on the drive rather than it actually being a failing drive?  It might be worth getting a diagnostics report (via the Tools tab)  and posting that here as that would show the SMART reports for the drives, and also the syslog for any error messages.

 

Actually doing what you suggest is not only possible, but quite easy.

 

If the disk has failed (red balled), unRAID will simulate the disk using parity and the other disks in the array. You can copy the files from the simulated disk (which will appear as if the real disk were there) to the newly added array disk. It might not be incredibly speedy, as parity and the destination disk are both in play for both the reading and writing phases, but it will copy your data off of the failed disk in a very straightforward manner. If the disk has not red balled yet, and is only showing signs of failure, the same method can be employed but it will go considerably faster. Once this is completed successfully, you can do a new config and exclude the failed or failing disk. Parity will rebuild.

 

What itimpi is talking about, I think, is using the new disk in the array to do a disk replacement. There is a way to even accomplish this with your disk in the array, but it would involve filling the new disk with binary zeroes and removing it from the array. This is tricky in situations where you have a failed disk, and I'd not recommend it over the method above.

 

I see you posted the diagnostics file but have not looked at that, but just wanted to weigh in on the question asked.

It is a good idea to NOT add disks to the array until very close to the time you might need them. That gives you the ability to use the new disk as either a replacement disk (for a true disk rebuild) or as a new disk. But once added to an array, using the disk as a replacement is not so easy.

 

 

[itimpi]

Hi

 

Here is the photo i posted from my other post:

 

 

I also attatched the Diagonostic report as an attatchment to this post.

 

Thank you very much for the help

The solution suggested by bjp999 is a good way to get the data off the problem disk before trying any recovery process.  At least once you have done that you will no longer be relying on parity to recover any data.  If you successfully copy the data onto another disk, do you intend to replace the problem disk, or to reduce the number of disks in the array?  The way forward varies according to what your intentions are at that point.

 

The SMART reports show that disk with serial WD-WCAZA1248672 has 7 pending sectors and 18 offline uncorrectable sectors.  Assuming I have followed the syslog correctly this is disk1 and that is the one reporting read problems in the syslog which is consistent with the SMART report.  The SMART reports for the other disks look fine.

 

The SMART values do not always indicate the disk is failing and it is possible that you could put the problem disk through a pre-clear and clear the pending sectors and offline uncorrectable error values.  However until that is done you do not want to trust that disk with any data.  These could be the first signs of imminent failure of the disk but the only way to find out is to test it thoroughly outside the array.

 

BTW:  The screenshot is not showing up when I view your posts, so do not know exactly what it is showing.

[SSD]

The solution suggested by bjp999 is a good way to get the data off the problem disk before trying any recovery process.  ...

 

After copying the data off onto the new array disk and reestablishing parity, I would say the "recovery process" is complete.

 

Once this is complete, there is additional testing that can be done on the apparently failed disk. I think that is what itimpi is referring to. This is optional, but may help restore some confidence in the disk to use it as a backup disk or some other purpose. Personally, when a disk starts to develop reallocated and pended sectors, I usually take it out of service.

 

[Nortrek]

The new disk i putted in yesterday is just zeroed over night, so if i coul`d remove that one from the array and put it in the slot where the disk is starting to show sign of errors.

 

The new disk i want to remove from the array got 0 bytes used so i don`t think it woul`d affect the parity disk much.

 

The disk that need to be changed is not redballed yet.

 

EDIT: Can i go into the failing disk 1 and copy the files there in to the new disk in the array, and then remove it from the raid ?

That woul`d be easyest so i don`t have to remove the new data disk to replace the failing one.

[trurl]

The new disk i putted in yesterday is just zeroed over night, so if i coul`d remove that one from the array and put it in the slot where the disk is starting to show sign of errors.

 

The new disk i want to remove from the array got 0 bytes used so i don`t think it woul`d affect the parity disk much.

 

The disk that need to be changed is not redballed yet.

 

EDIT: Can i go into the failing disk 1 and copy the files there in to the new disk in the array, and then remove it from the raid ?

That woul`d be easyest so i don`t have to remove the new data disk to replace the failing one.

If the new disk has been added to the array and it has been formatted, then it has affected parity because the new filesystem created when the disk was formatted actually wrote to the disk and parity. Zero bytes used (an empty filesystem) is not at all the same as a zeroed disk.

 

Not sure what you are saying in your EDIT but it sounds like you are maybe saying exactly what bjp999 suggested. Copy all of the data from the failing disk to the new disk. Then you can do a new configuration without the failing disk and let parity rebuild.

[Nortrek]

My English is not so good so i try to understand all what you wrote.

 

If i copy it over, can i use windows copy and just paste it from failing disk 1, to the fresh new disk 9 in the array, then new config and leave out the failing ?

 

It is very slow process to copy from the server via another computer and back to the new hardisk.

Woul`d be nice to have an filemanager on the server to copy internal.

 

Thank you so much for all this help