[DRBD-user] Two vice-versa services on two nodes/machines

Carson Gaspar carson at taltos.org
Sat Jul 29 23:59:29 CEST 2006

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


--On Saturday, July 29, 2006 9:15 PM +0100 Luciano Miguel Ferreira Rocha 
<strange at nsk.no-ip.org> wrote:

>> > On Sat, Jul 29, 2006 at 09:27:08PM +0200, Ondrej Jombik wrote:
>> >> - partition-A (service-A) will be master on machine1 and slave on
>> >> machine2 - partition-B (service-B) will be master on machine2 and
>> >> slave on machine1 Is this possible?
>> >> Does someone have this in production?
>
> You're distributing your load between two machines. That's a good thing
> to do. An otherwise idle server will be put to good use instead of just
> waiting for the other to fail.
>
> I see no drawback, only benefits, assuming that there's no convoluted
> dependency between them.

The main drawback is: what happens when one server fails? Can a single 
server handle the full load previously handled by 2 servers? You _must_ 
over-provision if you want services to stay fully functional in fail-over 
mode. If one server can't handle the load in active/standby, then it can't 
handle the load in active/failed. Now in many cases being up but slow is 
better than being down, but be _very_ sure you understand the risk decision 
you're making.

-- 
Carson




More information about the drbd-user mailing list