Quick question - I recently replaced my two SAS drives with SSD for my Cache.  As these are enterprise SSD, I went with Raid 0 (I know, still a risk).  Two 960GB drives in Raid 0.

 

I followed the process for moving the original Cache data to the array, replacing the original drives with the new drives, configured for Raid 0 via the Balance pulldown, then kicked off the move back to Cache.  All went and works fine.

 

Question: Why does the BTRFS filesystem for my Cache show Raid 1 for System and Metadata?  I did run the Perform Full Balance after the data was moved back as well.

 

 

 

 

Edited June 13, 20215 yr by mproberts

14 hours ago, mproberts said:

Why does the BTRFS filesystem for my Cache show Raid 1 for System and Metadata?

Because it's safer that way, metadata uses very little space comparatively and you'll have redundancy there, but if you really want you can force metadata to raid0 also using the CLI.

Thanks.  Is it a virtual Raid 1 on a Raid 0 formatted/pooled set of drives?  That's where my confusion lies and if this is some sort of virtual Raid 1, not clear how this would provide Raid 1 protection (1 drive failing).  

 

If my two drives are wholly a Raid 0 pool, how do I end up with a (reported) Raid 1 for metadata and system?

 

 

 

Btrfs has data and metadata chunks, all the data you copy there will use raid0, metadata (and system) is where the filesystem information is stored, only that part is raid1, if you for a example get a bad sector where metadata is stored you can lose the complete filesystem, with raid1 metadata there's still another copy if that happens, and again metadata takes so little space compared to data that it doesn't make sense to use raid0 for it, but you can if want.

Ok - that makes it clear.  Thank you!