Re: raid-on-raid during shutdown

[Alan Barrett <apb%cequrux.com@localhost>]

On Tue, 24 Mar 2009, der Mouse wrote:
> [problems with raidframe being marked dirty on shutdown, even a
> "clean" shutdown.]
>
> So, I've got two questions: (1) what am I missing here? and (2)
> if it's not obvious from the answer to (1), what is needed to
> make RAIDframe closes cascade properly, to get clean shutdowns of
> RAID-on-RAID setups?

I don't know why your case doesn't work, but it's probably just a bug, as
opposed to a design issue.

In the more general case, including arbitrary nesting of foo on
raid on foo on raid, there is a design issue.  I used to think
that a shutdownhook per file system or device would be suficient,
(e.g., instead of a shutdownhook for raidframe as a whole, have a
separate shutdownhook for each active raidframe device), but there
are pathological cases where even that doesn't work (last time this
discussion came up, you mentioned the case of a raidframe whose list
of components was changed while it was active).  To properly deal with
that, I think we need a new kind of shutdownhook that gets run in a loop
while still making progress.

--apb (Alan Barrett)