I have few questions regarding qtree, volume and raid group

1. Is it ture that raid group is added to a volume automatically?

Sort of, but that's not quite the way to look at it. Volumes have RAID group sizes associated with them. By default this is 14, but you can change it using a:

filer> vol options <volname> raidsize N

command if you so choose. N is the raid group size in disks.

New disks that are plugged into the system become hot spares, and can be subsequently added to any volume in the filer at any time. When disks are added into a volume, it will use them to "fill out" any not-full existing RAID group in the volume, up to the maximum RAID group size, at which point it will start a new RAID group automatically. Very simple.

2. is there any command for creating new raid group?

No. Such a command is not really needed.

3. when a qtree is created on a multi raid group volume, does the qtree

span

across all raid group?

Yes it does. WAFL manages the volume as a unit, and a qtree plus the data in it simply represent space within the volume.

do you have control on which raid group the qtree can reside on?

No, but then you shouldn't ever need such control. Trust in the magic! :-)

4. when the max disk per raid group is increased, does it affect all raid

groups?

As I mentioned above, RAID group size is defined on a per volume basis. If you decrease it, it does not effect existing RAID groups (RAID groups do not get shrunk down or anything like that). If you increase it, it does not effect existing RAID groups other than the "last" one, which may be partially full or indeed "full" by the definition of your previous, smaller RAID group size. New disks would get added to that last RAID group if you had increased the size.

5. I know if a new disk is added, it will be added to the last raid group

of

a volume.

Right, unless it's full etc...

Assuming the answer to Q4 is yes, then what happen if the last raid group of a volume is filled and there exists raid group(not the last one) in the same volume that is not filled?

<thinks>... <thinks>... :-)

Only the last RAID group gets new disks up to the new, larger RAID group size. We don't go back and try and fill-out the preceeding RAID groups to the new, larger RAID group size with new disks. The reasons to do something like that do not immediately spring to mind (other than trying to bring down your parity overhead as a percentage of total perhaps?).

Keith