On Tue, 20 Apr 2004, Bruce Momjian wrote:
I haven't yet seen the patch, but conceptually I'm conrned about how it
can handle situations like
this sequence
INSERT row X into database A
INSERT row X into database B
DBMIrror from A to B locks table
the mirror will fail.
>
> Where are we on this? Is it something that can be added to dbmirror as
> a patch?
>
> ---------------------------------------------------------------------------
>
> merino silva wrote:
> > Hi all,
> >
> > The method I've used to convert DBMirror to a
> > peer-to-peer replicator was two DBMirror instances
> > with one considered slave of other as the master.
> >
> > Here, I've stopped the loop back by dropping the
> > trigger when INSERT, DELETE, UPDATE through the
> > DBMirror and then creating the trigger again.
> >
> > The system was tested on the
> > ManddrakeLinux 9.2
> > Posgresql 7.3.4
> > and Perl v5.8.1
> >
> > A 'LOCK TABLE <table name> IN EXCLUSIVE MODE'
> > was used before dropping the trigger to prevent
> > updating the client database, because updates for that
> > database wont trigger to the remaining one while the
> > trigger is dropped.
> >
> > The modified DBMirror.pl is attached with this.
> >
> > Is this an acceptable solution for peer-to-peer
> > replication?
> > What can be go wrong with this approach?
> >
> > Please reply me with your suggestions.
> >
> > regards,
> > merino silva.
> >
> > __________________________________
> > Do you Yahoo!?
> > Get better spam protection with Yahoo! Mail.
> > http://antispam.yahoo.com/tools
>
> Content-Description: DBMirror.pl
>
> [ Attachment, skipping... ]
>
> > _______________________________________________
> > Pgreplication-general mailing list
> > Pgreplication-general@gborg.postgresql.org
> > http://gborg.postgresql.org/mailman/listinfo/pgreplication-general
>
>
--
Steven Singer ssinger@navtechinc.com
Dispatch Systems Phone: 519-747-1170 ext 282
Navtech Systems Support Inc. AFTN: CYYZXNSX SITA: YYZNSCR
Waterloo, Ontario ARINC: YKFNSCR