Actually, I don't plan to use it the way I suggested. While setting up a
raid system, I realized that it would be more convenient to create
single-disk mirrors, move the contents, and then add the original disk
back to the mirror on the fly (rather than having a third disk,
initializing the array on the first two, copying the third to the new md,
and removing the third disk from the system).
(And yes, I admit that there could be problems with the file contents
changing under the situation I described, but the backup would at least
back up the files that were there instead of complaining that they
changed out from under it.)
Jorj
On Mon, Nov 02, 1998 at 10:39:36AM -0800, Matthew Jacob wrote:
>
> Hot plugging would be good, but a rotating mirrors backup (this
> is similar to what Auspex has been known to do) really
> doesn't solve your static/consistent locking problem.
>
> You need to perform locking at the file level, if not block level.
> A lot of backup packages will do this, with greater or lesser
> success.
>
> On Mon, 2 Nov 1998, Jorj Bauer wrote:
>
> > (My apologies if this is already possible and I've missed it, but...)
> >
> > Are there any plans for the ability to add and remove mirrors on the fly?
> >
> > I can think of one particular situation where this would be exceptionally
> > useful. On a busy mail machine, it's difficult to get a static backup of
> > the contents of /var/spool/mail. If the raid tools supported the ability
> > to add and remove mirrors on the fly, one could have a redundant (third)
> > mirror which is detached nightly, mounted as a standalone filesystem
> > elsewhere, backed up, unmounted, and reattached to the /var/spool/mail
> > mirror.
> >
> > The rest of the raid package seems fairly thorough; this is the only
> > serious feature that I see missing. Or that I just don't see.
> >
> > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> > Jorj Bauer | [EMAIL PROTECTED]
> > Systems Programmer / Network Specialist | 200 S. 33rd St.
> > School of Engineering and Applied Science | Moore Building, Room 164
> > University of Pennsylvania | Philadelphia, PA 19104
> > http://binky.seas.upenn.edu/~jorj | O: 215/898-0575 F: 215/898-1195
> > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> >
