[Gluster-users] migration operations: Stopping a migration

[Brian Candler]

On Wed, Sep 05, 2012 at 10:40:58PM -0700, Eric wrote:
>    FYI: While working on another project, I discovered that the file
>    system attributes had been restored to their original state (or at
>    least what I remember their original state to be):

I would be delighted if someone could write a document explaining how
gluster actually *works*, especially with regards to state stored on the
bricks.  It's all very well having a black box, even with someone as big as
Red Hat behind it, but when the black box breaks it doesn't help you decide
the correct course of action to fix it.

Replicated volumes are the most important, followed by distributed volumes
(e.g.  I had a two-brick distributed volume which became unbalanced and I
couldn't see why)

Sorry I can't offer to write this document.  It's still a black box to me,
and what little documentation there is for gluster 2.x seems to be
irrelevant or even misleading for gluster 3.x.  I suspect things have been
changing in the 3.x series too.

Regards,

Brian.