Data on BTRFS Raid?

[*FordPrefect*]

Hi,

I am currently sorting out options for a home server project and often get the recommendation to use unraid.

However, I am not really comfy with what looks like a JBOD with Parity Disk approach. I saw that one can format the data disks to BTRFS and that for the unraid cachepool one can even use BTRFS Raid1 with scrubbing and self healing functionalities. But I didn't see anything about the option to use this functionalities with the datapool.

So I wonder if I just overlooked it and if not, whether you have plans to change it. I have the feeling that the checksums of BTRFS and the self healing scrubbings are superior to the default unraid approach, but maybe I am wrong as I am certainly not an expert.

 

Best regards.

[Vr2Io]

Unraid have 2 storage pool offical. Array pool and cache pool.

 

Array pool have file pooling function ( not JBOD ), so user would see all file in same volume although file actually in different disks. Optionally, parity disk can add for disk rebuild in case disk failure, but parity won't know anything about filesystem or file.

 

Due to too much layer in array pool, performance won't good, so cache pool mainly use to overcome this.

 

In fact, both pool could be BTRFS or other file system ( array pool could mix of different filesystem disks but not allow any RAID function ).

Edited December 16, 2019 by Benson

[JorgeB]

Some of the main advantages of Unraid are the ability to use different size devices on the array and in case of more disks failures than parity can handle every disk can always be read individually on any Linux distro, so because of this all array data devices use an individual file system, you can still use btrfs to detect data corruption but it can't self heal since there's no redundancy, you'd need to restore from backup, cache pool can have up to 24 devices and for that you can use a redundant self-healing btrfs pool.

[*FordPrefect*]

Thanks for clarification. I see that it's similar to snapraid regarding this aspect. I guess flexibility to configure a storage pool differently wouldn't harm. But I don't mind. I guess I will go another route.

[BRiT]

10 hours ago, johnnie.black said:

Some of the main advantages of Unraid are the ability to use different size devices on the array and in case of more disks failures than parity can handle every disk can always be read individually on any Linux distro, so because of this all array data devices use an individual file system, you can still use btrfs to detect data corruption but it can't self heal since there's no redundancy, you'd need to restore from backup, cache pool can have up to 24 devices and for that you can use a redundant self-healing btrfs pool.

 

I don't suppose you've played around with duplicate data setup of BTRFS on a single drive? Is that even a thing, I know there's options for the metadata just not certain if BTRFS can mirror drive data to itself.

[JorgeB]

10 hours ago, BRiT said:

I don't suppose you've played around with duplicate data setup of BTRFS on a single drive? Is that even a thing, I know there's options for the metadata just not certain if BTRFS can mirror drive data to itself.

You can use DUP data profile, all data will be duplicated on the same drive, but never used it since not an use case for me.