Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Steve Purkis wrote:
> Hi list,
>
> I'm new to DRDB - in fact I'm just downloading it to give it a run on
> 2.4.27 at the mo. I've got a question that still is unclear to me after
> reading through the docs on drdb.org; it may be a stupid question;
> apologies if it's already been asked (I googled but couldn't find
> anything):
>
> How does DRDB do mirroring?
>
> I realize the site says that "you could see it as a network raid-1". I
> guess my question is a bit more specific: does that mean it really does
> the same job that md does? Or does drdb try and sync only blocks that
> are out of sync (as Peter Breuer's "Fast RAID-1" does). Or does it do
> something else? So I guess I could rephrase the question:
>
> Does DRDB do anything to try and avoid re-syncing an entire device
> to the remote machine?
>
> In the long run, I'm thinking of the scenario where, in a Linux-HA
> cluster, the primary goes down, services failover to the secondary, and
> then the original primary comes back up. At this point, it makes sense
> to only sync the differences between the two drives to avoid sending
> potentially gigabytes of data over the net.
>
> I guess the next question is:
>
> If DRDB doesn't do any 'intelligent' resync, are there any plans to?
That's easy.
It does intelligent resync. In fact, it will never do a full resync of two
nodes (except the first time).
In fact, I believe it does a better job of doing intelligent resync than
"Fast RAID-1". For details, see this article:
http://www.drbd.org/fileadmin/drbd/publications/drbd-activity-logging_v6.pdf
--
Alan Robertson <alanr at unix.sh>
"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce