[DRBD-user] taking a drbd device back to md

Mark Watts m.watts at eris.qinetiq.com
Wed Feb 3 13:31:58 CET 2010

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Wed, 2010-02-03 at 04:24 -0800, Yan Seiner wrote:
> Mike Lovell wrote:
> > Yan Seiner wrote:
> >> I have a drbd device that I want to convert back to the underlying md 
> >> device.  Is there some way to do that in place?  Or do I have to move 
> >> all of the data off the device and reformat?
> >>
> >> And how do I reformat the device to remove all of the drbd info?
> >>
> >>
> > i don't think you have to do anything to be able to use the underlying 
> > md device except stopping drbd and unloading the kernel modules. once 
> > drbd is off, you should be able to mount and use any filesystems you 
> > have on the md device without any changes.
> >
> > i wouldn't bother trying to remove the drbd metadata because it is 
> > using a small portion of the disk. i haven't done it before and it 
> > sounds kind of risky. iirc, it is only a couple megabytes of space for 
> > hundreds of gigabytes of capacity. this is going to gain so little 
> > space back i wouldn't risk it. i would just stop drbd, remove the 
> > disks from drbd.conf, and use the device normally.
> 
> Exactly correct.  Being paranoid, I backed up the entire partition, then 
> rebooted the machine making sure drbd would not start.  I could access 
> the data with no problem.
> 
> This does beg a question, though.
> 
> I start drbd and then run fsck on the drbd device.  It would be faster 
> to run fsck on the underlying file system, and then start drbd.  That 
> would make life easier as one could do all the file operations before 
> even starting the network at boot.

Don't forget that DRBD is only replicating blocks - it doesn't care
about the actual data (ie: filesystem) in use.

This implies that if you have filesystem corruption on one node, you'll
have it on the other too. Net result is, if you fsck /dev/drbdX, you're
in effect fsck'ing the remote node too.

Always better to use the top-level block device in the stack to maintain
consistency.

Mark.

-- 
Mark Watts BSc RHCE MBCS
Senior Systems Engineer, Managed Services Manpower
www.QinetiQ.com
QinetiQ - Delivering customer-focused solutions
GPG Key: http://www.linux-corner.info/mwatts.gpg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20100203/a86c6998/attachment.pgp>


More information about the drbd-user mailing list