July 3, 201412 yr Hi - My server is still fairly new, so I've still got time to adjust before data gets spread all over the place. Right now I've got several user shares setup without any Included/Excluded disks specified or Split Levels set. My preference would be to fill up disks by user share, but I'd still like the cross disk advantages that user shares offer. In other words, I'd like writes to the Backup user share to be biased towards disk1 and only start writing to other disks is disk 1 fills up. I'd like writes to TV Shows user share to be biased towards disk2, etc. My rationale is that when a disk fails I'll know the sense of urgency about replacing it based on the content, plus in a worst case scenario if I need to move a disk to another machine to recover data it would be nice to have the data for a particular share largely clustered together rather than spread around. I haven't figured out how to do this yet, though. Any suggestions?
July 3, 201412 yr ...My rationale is that when a disk fails I'll know the sense of urgency about replacing it based on the content... Suggest you rethink this. The contents of a single failed drive are still accessible since unRAID will simulate it using the other drives plus parity. However, a single drive can only be rebuilt if all other drives plus parity are good. So, if you delay replacing a bad drive then another failure means you lose both.
July 3, 201412 yr Author Hmm, good point - I wasn't thinking that all the way through. Still, I wouldn't mind a solution to establishing a preference order for writes if anyone has one...
July 3, 201412 yr Hmm, good point - I wasn't thinking that all the way through. Still, I wouldn't mind a solution to establishing a preference order for writes if anyone has one... Think fault tolerance and not about what is located where on X disk. You have no clue which disk will go first and when one does you simple restore it on a new one. You aren't going to spend a month trying to figure out what was in that particular disk because you just restored it. Also if you were running RAID or ZFS or BTRS RAID 5/6 you would have no clue or way to see where anything is nor would you care. All you would care about is it has fault tolerance so my advice is to do the same with unRAID (using split levels if you want to get a little more specific).
July 4, 201412 yr Author Agreed, I am familiar with traditional data striping approaches to raid... But that's one of the reasons I chose unRAID;). In the final analysis I'll be fine if I can't specify data locations, I'm just doing a little OCD tinkering as I move into my new home . Thanks
Archived
This topic is now archived and is closed to further replies.