How to Balance Disk Usage?

[interwebtech]

Back when I consolidated 2 uR servers into one, I didn't leave much slack space on each drive (trying to pack as much as possible at that time). Most are at 99% full. I just upgraded 2 drives to new double capacity drives and want to move some folders to the larger drives to bring all the full drives to @90% full. 2TB drives means I just need to move 1 movie (2 GB) to accomplish that

 

I have 1 huge share named Movies1 than spans all drives in the array. When I opened MC to move 1 or 2 folders to the larger drives off a full disk, I noticed that both sides of MC had the same folder name. Hmmm... that might be a problem. I have read about losing data like this so backed off and came here to ask how to pull this off.

[kaiguy]

Instead of moving from a user share to a disk share, why not just go into individual disk shares and move to another disk share?

 

So, instead of opening up MC and going to /mnt/user/Movies1, go to /mnt/disk1 on the left, /mnt/diskXX/Movies1 on the right (where XX is the one of the new drives) and some of the movie directories over.

 

Also, unless my math is off, 90% of 2TB would be 1800 GB, so you'll be needing to move closer to 200 GB of movies, not 2 GB.  (When rocking moves this large via telnet/ssh, you'll want to use screen so you don't have to stay connected the whole time).

[interwebtech]

Instead of moving from a user share to a disk share, why not just go into individual disk shares and move to another disk share?

 

So, instead of opening up MC and going to /mnt/user/Movies1, go to /mnt/disk1 on the left, /mnt/diskXX/Movies1 on the right (where XX is the one of the new drives) and some of the movie directories over.

 

Also, unless my math is off, 90% of 2TB would be 1800 GB, so you'll be needing to move closer to 200 GB of movies, not 2 GB.  (When rocking moves this large via telnet/ssh, you'll want to use screen so you don't have to stay connected the whole time).

 

oops on the math lol.  200 GB sound better.

 

On both source and destination disk, the same movie folder are inside a top level folder named Movies1. Looking at /mnt/disk3/Movies1/ on the left and /mnt/disk11/Movies1/ I am seeing the same folder name e.g., /127_hours/ at the top of both list and peeking inside each I see that somehow this movie got spanned across the 2 disks. One side holds the video & data files (XML etc) and the other has just the cover image.

 

So I am guessing its safe to move the files from disk3 to disk11 and then just delete the empty dir on disk3?

 

[kaiguy]

Yes, as long as you're working with disk shares on both sides, it's no problem copying to one and deleting the empty directory.  It's just the way it ended up getting distributed due to your split level.

 

[skyhawk]

 

oops on the math lol.  200 GB sound better.

 

On both source and destination disk, the same movie folder are inside a top level folder named Movies1. Looking at /mnt/disk3/Movies1/ on the left and /mnt/disk11/Movies1/ I am seeing the same folder name e.g., /127_hours/ at the top of both list and peeking inside each I see that somehow this movie got spanned across the 2 disks. One side holds the video & data files (XML etc) and the other has just the cover image.

 

So I am guessing its safe to move the files from disk3 to disk11 and then just delete the empty dir on disk3?

 

You can prevent this through the disk spanning options under your Movie Share. I really dont have much experience with it, so I wont tell you what to do, but read the wiki and there is an option that will keep all subfolders grouped together. Its not hard, I just dont want to give you the wrong option. But, this wont fix whats already split.. it will just prevent splits in the future.

 

Move using the MC/midnight commander. Its fairly simple/stupid and its the fastest option.

 

From what Ive read, it will write to parity as you move it, so you dont need to run another parity check right away. But... if moving a lot of data, it might be a good idea to let one run tonight just for peace of mind.

[pkn]

... On both source and destination disk, the same movie folder are inside a top level folder named Movies1. Looking at /mnt/disk3/Movies1/ on the left and /mnt/disk11/Movies1/ I am seeing the same folder name e.g., /127_hours/ at the top of both list and peeking inside each I see that somehow this movie got spanned across the 2 disks. One side holds the video & data files (XML etc) and the other has just the cover image.

This happened to me when I accidentally copied movie files into user share /mnt/user/Movies1/127_hours, instead of using direct access to particular disk /mnt/disk3/Movies1/127_hours . unRAID decided himself which particular file goes where.

 

I guess the same happens when the Mover script works, but I don't use Mover, so I can't be sure.

[interwebtech]

Thanks for the suggestions. I have what I need now

[garycase]

...  Most are at 99% full. I just upgraded 2 drives to new double capacity drives and want to move some folders to the larger drives to bring all the full drives to @90% ...

 

Why?  There's NO reason to waste 10% of your disk, especially if the content is effectively static (i.e. you're not deleting/replacing files on the full disk).   

 

 

...  I have read about losing data like this so backed off and came here to ask how to pull this off.

 

The data loss issue is only when moving using the user shares ... as long as you access everything via the disk# you'll be fine.    e.g. moving from  /mnt/disk3/Movies1/  to  /mnt/disk11/Movies1/  is fine; just do NOT use the user share reference [i.e. /mnt/Movies1/ ]

 

 

[BobPhoenix]

The only bug free way on a user share is if you are moving to different directories (from Movie1 to Movie2 for instance) or if you move from a cache drive to user0 share which doesn't include the cache drive in the share - which is how the mover script moves files to the array from the cache drive.  But it is real easy to get user and user0 mixed up so I wouldn't use it - found that out the hard way myself. I would just go disk to disk or cache to disk and avoid the bug completely.