hypothetically if I have a server with say 10TB of free space and I have a drive failure, can I just remove the drive and rebuild parity with less space?

 

When building my server, would I be better off building with less overhead?  Meaning, say I buy 10TB worth of drives, would it benefit me to only install what I need and leave the rest for when I get close to full?

 

thanks

Yes, but you must set a new disk configuration and calculate parity on the.neq cnfig.

hypothetically if I have a server with say 10TB of free space and I have a drive failure, can I just remove the drive and rebuild parity with less space?

 

When building my server, would I be better off building with less overhead?  Meaning, say I buy 10TB worth of drives, would it benefit me to only install what I need and leave the rest for when I get close to full?

 

thanks

 

As Joe said, it's fairly simple to pull drives and recalculate parity protection with only the remaining data. The issue with that scenario is that you are risking another drive failure with no protection until a parity build and check are completed successfully. You would be better served to only utilize the drives you need, that way if a single drive fails, you will have spares waiting to replace any failures right away, without losing data. Another major issue for most people is that they don't know exactly what data files are on any given disk, because unraid shares can span several disks. It could be a major pain to figure out exactly what has gone missing when you remove an arbitrary disks, unless you only use disk shares, and keep inventory on what files occupy each disk.

 

 

As Joe said, it's fairly simple to pull drives and recalculate parity protection with only the remaining data. The issue with that scenario is that you are risking another drive failure with no protection until a parity build and check are completed successfully. You would be better served to only utilize the drives you need, that way if a single drive fails, you will have spares waiting to replace any failures right away, without losing data. Another major issue for most people is that they don't know exactly what data files are on any given disk, because unraid shares can span several disks. It could be a major pain to figure out exactly what has gone missing when you remove an arbitrary disks, unless you only use disk shares, and keep inventory on what files occupy each disk.

 

Oh wow!  I guess I always thought because it was parity protected that I could rebuild with one less drive (as long as the total array had enough space).  It would have never ocurred to me that I would lose data by removing a failed drive and rebuilding parity.  I will take your advice and only build with 1-2TB overhead and expand from there.  I had 9.5TB worth of storage ready to install for only 2TB of data 

 

pleas correct me if i am missing something here....

You can indeed rebuild with one less drive if you have enough space, but it's far from simple. In a nutshell, you would copy all the data from the parity simulated failed drive over to one or more of the real drives, then you can either do as Joe said and remove the failed drive and rebuild parity, because there would be no data left on the failed drive. Or, in a much more complicated maneuver (which I pulled off several months ago) you can zero out the newly emptied parity simulated failed drive, and then tell the server to trust that the parity is indeed still correct, because you reverted the failed drive to all zeros, which means the parity will still be correct when you remove the drive. Either way if a second drive fails before the procedure is completed you will lose whatever data was on both drives.

 

Repeat after me, unRaid is not a backup if it holds the only copy of your data.

 

It's a dandy tool for protecting your data against a single disk drive failure, and is a GREAT place to store backups of data that exists elsewhere for convenience sake, such as rips of an extensive DVD/BluRay movie collection. It doesn't offer protection against accidental deletion, or user induced corruption of data.

It sounds like a safer play would be to build with less storage overhead and replace or expand as needed.

 

Thanks for everyones input

It sounds like a safer play would be to build with less storage overhead and replace or expand as needed.

 

Thanks for everyones input

 

Yep, that is exactly what I do.

 

I have 16x precleared 3TB drives in my unraid all in hot swap caddies.

But... I only pushed in 4 drives at first. the rest were sitting in the bays, but not plugged in (with the handle closed, the lock tab was holding them out).

 

As the unraid get down to about 2 TB free, i would stop the array and push the next drive in, assign it and restart the expanded array.

I am now up to 10 drives installed. I am hoping the last 6 outlast this drive shortage.

 

this saves power and wear and tear on the drives that i am not using.

remember, un-assigned drives in an unraid box do not spin-down. you're better off not plugging it in.