I could not figure out from https://lime-technology.com/network-attached-storage/ how the core of unRaid (eg: without the cache) is different from Raid5 (or Raid6 now it supports two parity disks in version 6). I would suggest you rewrite that page to raise more interest in your product like this page did:

 

https://www.anandtech.com/show/9463/lime-technologys-unraid-6-brings-containers-and-virtualization-to-nas-units 

 

• Avoiding striping ensures that it is trivial to take out a disk, mount it on another Linux system and copy off its contents. To drive home the advantage of this aspect - in case of simultaneous failure of two or more disks, it is possible to recover at least some data from the array by mounting the remaining good disks on another PC. (In the case of a RAID-5 array, the whole data is toast).
• Avoiding striping ensures that only the relevant disk needs to be spun up to read or write data. This may result in substantial power savings for multi-bay units where the power consumption of the member disks far outweighs the consumption of the system components.

 

Both of these features where exctly what i am looking for.

 

Then i failed again finding good details exactly how  that "desaster recovery" works. I gave up on wiki pages mentioning NTFS (did unRaid ever used NTFS ?) and reiserFS and thought it might be easier to ask here:

 

- When i take a disk out of the unRaid hook into a linux system, i can mount it (read-only) - as long as i have XFS ? 

- How much of my original  directory structure and file names will i find again , how much obfuscation does unRaid add ? multiple partitions, etc.. ?

- When setting up the unraid, how can i control which disk a file/directory goes into ? (i guess i can create 5 group/volumes/users? one for each of eg: 5 data disks... ?) 

 

Last but not least: How will that desaster recovery work with encrypted disks, e.g. step by step (LUKS etc.) for a non unRaid linux system would be great.

 

In 2005 i lost a hardware raid system because it was all proprietary and i couldn't perform desaster recovery. Since then i've been running with linux but without goo/simple redundancy. In theory, unRaid looks exactly what i am looking for. Give or take the answers to above questions.

 

Thank you very much

Have a read here.

https://lime-technology.com/wiki/Parity#How_parity_works

See if that answers your questions on recovery.

Each disk is a totally separate filesystem, the only link between disks is the logical concatenation of all root folders on all array and cache disks to form the user shares. You can address them individually or as user share groupings. You could have some disks as XFS, some as BTRFS, some encrypted, some not, and they all will work together.

1 hour ago, te36 said:

- When i take a disk out of the unRaid hook into a linux system, i can mount it (read-only) - as long as i have XFS ? 

 

Yes as long as you have the same file system.  Usually this is XFS but you also have option of btrfs or ReiserFS (though no longer recommend to use that).  In the case of a btrfs multi-device cache pool, you may need to move all devices to the other system.

 

1 hour ago, te36 said:

- How much of my original  directory structure and file names will i find again , how much obfuscation does unRaid add ? multiple partitions, etc.. ?

 

All of it.  None of your data is "obfuscated" in any way.  Devices formatting under unRAID OS only have a single partition that spans the entire device.

 

1 hour ago, te36 said:

- When setting up the unraid, how can i control which disk a file/directory goes into ? (i guess i can create 5 group/volumes/users? one for each of eg: 5 data disks... ?) 

 

This is done on a per-share basis.

 

1 hour ago, te36 said:

Last but not least: How will that desaster recovery work with encrypted disks, e.g. step by step (LUKS etc.) for a non unRaid linux system would be great.

 

You just use 'cryptsetup' like any other LUKS volume.  Of course you need to know the passphrase.

8 hours ago, jonathanm said:

Have a read here.

https://lime-technology.com/wiki/Parity#How_parity_works

See if that answers your questions on recovery.

 

Thanks. I actually didn't ask about the parity because i understand it.

Your URL does not describe though which algorithm is used for the second parity disk.

 

7 hours ago, limetech said:

This is done on a per-share basis.

 

Thanks! So each disks would have at the top level one directory for each share (probably with the name of the share) and all files for that share would be inside that directory ? 

 

Bonus questions for same bucket (reliability);

 

I never had the time to move to XFS, primarily because its so flexible and complex. It seems to have some in-file-system protection against bit rot (checksums per file ?). Does unRaid provide gui to run this type of checking periodically ? 

 

 

1 hour ago, te36 said:

 

Thanks. I actually didn't ask about the parity because i understand it.

Your URL does not describe though which algorithm is used for the second parity disk.

 

 

Thanks! So each disks would have at the top level one directory for each share (probably with the name of the share) and all files for that share would be inside that directory ? 

 

Bonus questions for same bucket (reliability);

 

I never had the time to move to XFS, primarily because its so flexible and complex. It seems to have some in-file-system protection against bit rot (checksums per file ?). Does unRaid provide gui to run this type of checking periodically ? 

 

 

 

Each disk may have one or more directories representing one or more shares.  If a share spans multiple disks, then each of the disks used by that share will have a directory identically named for that share.  Mulitple shares can span the same multiple disks at the same time if needed.  As noted above, the allocation of disks to shares is on a per share basis - the configration page for a share allows you to either include or exclude specific disks for a share or simply to use all available disks.  You then have additional control of the method used to allocate files within a share across disks as they are added. 

 

There's a lot of info here...

 

http://lime-technology.com/wiki/UnRAID_Manual_6#User_Shares_2

 

But ultimately I'd recommend getting a trial icense and having a play.  Much better than just reading.

 

There are file integity checkers you can apply, but at the basic level, unRAID natively allows you easily to schedule a regular parity check (typically monthly or weekly, but up to you).

Thank you all. Yes. certainly at the level of interest now to try it out. Just need to figure out a system to do it on. Don't want to start with one of my main servers now, but need to set up some 3 disk playground in my infinite nonexisting spare time *sigh*.

 

Btw: Seeing how you have a main and backup server: Anything recommend by unRaid to synchronize two servers ? In my case the main issue is that i like to move stuff around the directory trees and rename it over the years - to better organize the content over time, and given how the two servers are located on two tectonic plates, the bandwidth between them really requires to not have to copy in such cases but to just sync movements/renamings. Kinda possible to script with rsync manually but difficult. And didn't have the time to investigate other options. Most "cloud sync/storage" software i saw was way too obfuscated and not really focussed on such a job. A solution like unRaid trying to sell licenses for many server might have more business interest in supporting this. So any feature in unRaid for this would be icing on the cake.

 

To backup between servers I use some simple rsync scripts.  But there is no single best solution, everyone may have different needs.  I run my backups manually, when I have the need - the rest of the time the backup server is physically switched off (hence nothing can write / modify / delete content at other times).  Using rsync means that I only copy across files that have been updated on the primary server, but as I have it the whole file gets copied.  Maybe there's some way to copy only the differnces, but I've not looked at that (my servers are very much on the same tectonic plate).  rsync also allows a comparison of files on each server (including file contents) but without the file itself being copied acros the link.  I sometimes use that to verify the backup, although I am currently learning how to use the Dynamix File Integrity plugin.

Thanks. That sounds similar to what i (hate to) do. You can trick rsync by creating a symlink tree copy after each rsync copy,  so when you start moving/renaming files/directories, rsync can be made to copy based on inodes. But its tricky and doesn't always work. There is also in some branch a file based source/receiver mapping tracking... which also doesn't work

Edited February 1, 20188 yr by te36

Oh, one other thing coming to brainstorming with myself how to adopt unraid: I had over the years a few times the problems that the hardware used required kernel patches. For example now i still have a 5*8TB external disk subsystem  for backup that has a SATA port-multiplier that requires a patch for my server motherboard that never got into linux mainline.

 

Is there a way to patch up unRaids kernel/modules ?