[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Master and Slave (and the other way around) [auf Viren überprüft]



Hans Moser <hans.moser@ofd-sth.niedersachsen.de> writes:

> Kevin Spicer schrieb das Folgende am 19.11.2004 10:36:
>
>> I'm just in the process of doing this myself, and it works fine in my
>> test setup.
> Ok, sounds great. Thank you - and Dieter.
>
>> Your configuration below isn't quite correct though.  You
>> need to put the central (superior) database at the end of the database
>> definitions, you also need to add the 'subordinate' directive to each of
>> the other databases.
> I'll try. I just pasted together parts of the manual to illustrate my
> plan. Thanks for advice.
>
>> I don't know about slurpd replication though, I'm using syncrepl for
>> replication.
> Anyone?
>
> slurp vs. syncrepl:
> Are there the same (or other) restrictions for a syncrepl-slave as for
> a slurp-slave or are there at least just advantages?
> IHMO both are "one-way-replication"-systems master -> slave. The
> servers are not equivalent: server1 <-> server2. Am I right?

Replication is allways a master --> slave direction. server1<->server2
would be a multimaster environment.
For the time beeing I would prefer slurpd to syncrepl,in particular
persistant state raised some problems in the past. But I use
refreshOnly on my notebook to update the database when enver it is
connected to my local network.

-Dieter

-- 
Dieter Klünter | Systemberatung
http://www.dkluenter.de
GPG Key ID:01443B53