I have a number of older 1tb drives that I'm thinking about added to my server.  I don't need them in my capacity but I thought I would use them until they die.  Once they do die can I remove them and have unRAID rebuild the array (ie, I won't replace them)?

 

example:

4 x 2tb (giving 6tb of storage)

3tb of data on the drives

add 2 x 1tb drives.

total of 8tb of storage but still only 3tb of data.

 

oldest 1tb drive dies.

Can I remove it and have unRAID build my array back to five drives and 7tb of storage?

 

I'd like to run these old drives into the ground and maybe prolong the life of my newer drives.

 

Thanks.

 

When the drive dies and you don't replace it, you will lose everything that is on that drive.  Your parity will be invalid and cannot be used to rebuilt any other drive that fails.  Basically, once you put a drive into an array, it must be there forever.  If any drive drive fails, you must replace it with a drive of equal or larger capacity. 

 

By the way, I have never heard of any type of RAID system that will allow you to reduce the number of drives.  I know of none (with the exception of unRAID) that will allow to add another drive after the array is defined.  In fact, most of them require that any failed drive be replaced with a drive  that is identical in capacity.

I have a number of older 1tb drives that I'm thinking about added to my server.  I don't need them in my capacity but I thought I would use them until they die.

Very bad idea. If one of your "good" drives fails, it requires perfect performance from every other drive to rebuild the missing drive. Knowingly using a marginal drive is asking for data loss.

Basically, once you put a drive into an array, it must be there forever.  If any drive drive fails, you must replace it with a drive of equal or larger capacity. 

 

By the way, I have never heard of any type of RAID system that will allow you to reduce the number of drives.  I know of none (with the exception of unRAID) that will allow to add another drive after the array is defined.  In fact, most of them require that any failed drive be replaced with a drive  that is identical in capacity.

 

This is not true. Use the New Config button in Version 5 or enter "initconfig" on the command line after removing a drive and parity will be generated for the remaining drives. The data on the missing drive will be gone unless it is copied first.

In /theory/ you could bring the array up with the drive failed...copy the data onto a drive that's still working (by using the rebuilt data to do the copy, same way you "see" a failed drives data in a user share because it is constructed on the fly from parity) then when the data is safe on the good drive do a "new config" and start a fresh parity run.

 

In reality?  that leaves you unprotected for a long time where you are using the disks a LOT...so it is a high risk operation IMHO

In /theory/ you could bring the array up with the drive failed...copy the data onto a drive that's still working (by using the rebuilt data to do the copy, same way you "see" a failed drives data in a user share because it is constructed on the fly from parity) then when the data is safe on the good drive do a "new config" and start a fresh parity run.

 

In reality?  that leaves you unprotected for a long time where you are using the disks a LOT...so it is a high risk operation IMHO

 

This is typically the procedure that people use.

In /theory/ you could bring the array up with the drive failed...copy the data onto a drive that's still working (by using the rebuilt data to do the copy, same way you "see" a failed drives data in a user share because it is constructed on the fly from parity) then when the data is safe on the good drive do a "new config" and start a fresh parity run.

 

In reality?  that leaves you unprotected for a long time where you are using the disks a LOT...so it is a high risk operation IMHO

It is exactly the procedure people use.  Specifically, parity will be re-established as soon as the new calcs are completed.    Frequently the time without parity this way is far less than the time waiting for a replacement disk through an RMA.

 

This is typically the procedure that people use.