[DRBD-user] Changes from lower level block device

Felix Frank ff at mpexnet.de
Tue Feb 7 09:16:10 CET 2012

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


Hi,

On 02/07/2012 08:36 AM, Jaap78 wrote:
> 
> Hello,
> 
> Normally DRBD writes changes from device /dev/drbd[x] to a lower level block
> device (for example /dev/sda) and then submit it to the secondary node.
> 
> Does DRBD also submit changes to the secondary node, if the change came from
> the lower level block device (/dev/sda)?

no, it cannot possibly do that. It's crucial that the backing device is
not touched once DRBD is operational. Otherwise you won't ever find out
what blocks have diverged by writing through this "bypass".

> I am looking for a way to migrate a 8TB SAN to SAN storage server.

So your idea is to add DRBD metadata to your existing SAN and have any
changes replicated from then on?

I fear there is no way to do this. You will need a downtime, but it
should be possible to keep it as short as a couple of minutes:
1. add DRBD as you seem to be planning
2. shut down your services, unmount
3. do drbdadm create-md, go primary and bring the drbd up
4. mount (drbd instead of backing storage)

Your downtime comprises step 3, and can be done rather quickly (a bit of
practice doesn't hurt for this).

This way you opt to set up a standalone primary at first. You can then
take your time for setting up the peer. If you need to keep the initial
sync short as well, some more planning is in order, so that ideally you
can do some variation of truck based replication.

HTH,
Felix



More information about the drbd-user mailing list