[DRBD-user] drbd and lvm

Lars Ellenberg Lars.Ellenberg at linbit.com
Fri Mar 19 22:28:19 CET 2004


/ 2004-03-19 20:07:21 +0100
\ Baldur Norddahl:
> > > My hope is then if we get additional disks in the servers,
> > > that I can use LVM to expand the DRBD partition and have
> > > DRBD mirror the larger partition. All without destroying the
> > > filesystem, and preferably without downtime.
> > > 
> > > Will this work?
> > 
> > Yes. I put the link on http://www.drbd.org/faq.html
> > With 0.6. you will have downtime.
> > With 0.7 you currently won't have a stable system :)
> 
> Ok will you be able to upgrade from 0.6 without downtime (except for the time a
> failover takes that is)? Just curious, a short planned downtime would probably
> be ok for this project.

Hm. *unless* someone writes a compatibility mode...
most likely this won't happen.

So you need to plan a downtime to upgrade 0.6 -> 0.7 later this year.

If you plan to deploy some new 0.6 based cluster now, and
expect to upgrade once 0.7 is settled, you should prepare
for the persistent bitmap and metadata.

0.7 needs 128MB stable storage for persistant bitmap and metadata
per device, either at the end of the lower level device, or within
some other dedicated block device.

You can have all your DRBDs share one dedicated partition for the
metadata, provided it is large enough.

	Lars Ellenberg



More information about the drbd-user mailing list