[Ocfs2-tools-devel] [PATCH] Option to skip cluster checking in tunefs.ocfs2

[Lars Marowsky-Bree]

On 2009-10-17T04:12:09, Joel Becker <Joel.Becker at oracle.com> wrote:

> > That's correct. (Remember the LVB size issue?)
> 	I'm not sure what you mean.  Are you saying that SLES isn't
> going to take the fix for the LVB size issue?

We'll take it, but it breaks the ABI, if I'm not mistaken. And the last
I knew the fix wasn't done yet (or maybe I missed it in the commit
logs?) So we need something for the present.

> > It also seemed impossible to alter the fs while umounted and with the
> > cluster stack wholly stopped, which seemed like a desirable feature.
> 	It's not desirable, because you don't know that it is stopped
> elsewhere.  The only way to know is to have the stack up and check.
> 	Essentially, we make every effort to ensure that we don't lose
> people's data.  This includes making it hard for them to shoot
> themselves in the foot.  Now, there are places we can't avoid it.  But
> we try to restrict ourselves to only those places.

I'm perfectly fine with protecting admins. I'm also perfectly fine with
there being override functions in case something goes wrong, or when the
admin wants to do something while the stack is, in fact, down. (Perhaps
fix something which would cause the stack to crash? ;-)

> 	There are a couple of things that cannot be done with the
> cluster active.  In tunefs.ocfs2, we provide specific options for those.
> Thus, only those can be invoked in a non-cluster-coherent manner.

That makes sense. The override seems a work-around in the mean-time
though.


Regards,
    Lars

-- 
Architect Storage/HA, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde