lxd/storage/drivers/driver/btrfs/volumes: Enable nodatacow on block r…

…aw disk files

Enable nodatacow so that random writes don't cause fragmentation and old extents to be kept.
BTRFS extents are immutable so when blocks are written they end up in new extents and the old
ones remains until all of its data is dereferenced or rewritten. These old extents are counted
in the quota, and so leaving CoW enabled can cause the BTRFS subvolume quota to be reached even
if the block volume file isn't full.

Fixes canonical#9124

Signed-off-by: Thomas Parrott <thomas.parrott@canonical.com>

lxd/storage/drivers/driver_btrfs_volumes.go

@@ -52,6 +52,23 @@ func (d *btrfs) CreateVolume(vol Volume, filler *VolumeFiller, op *operations.Op
 		if err != nil {
 			return err
 		}
+
+		// Create empty file so we can set the compression property on it.
+		f, err := os.OpenFile(rootBlockPath, os.O_RDWR|os.O_CREATE, 0600)
+		if err != nil {
+			return errors.Wrapf(err, "Failed to open %s", rootBlockPath)
+		}
+		f.Close()
+
+		// Enable nodatacow so that random writes don't cause fragmentation and old extents to be kept.
+		// BTRFS extents are immutable so when blocks are written they end up in new extents and the old
+		// ones remains until all of its data is dereferenced or rewritten. These old extents are counted
+		// in the quota, and so leaving CoW enabled can cause the BTRFS subvolume quota to be reached even
+		// if the block volume file isn't full.
+		_, err = shared.RunCommand("chattr", "+C", rootBlockPath)
+		if err != nil {
+			return fmt.Errorf("Failed setting nodatacow on %q: %w", rootBlockPath, err)
+		}
 	}
 
 	err = d.runFiller(vol, rootBlockPath, filler)