[Bucardo-general] Replication aborted because of DDL change
akpgeek at gmail.com
Wed Feb 17 15:41:05 UTC 2010
Thanks for the repsonse Greg. One question/clarification. Since the
replication process is stopped now. If I sync the master / slave with the
DDL change and start the Bucardo_CTL, will it take care of the data for the
rest of the tables or I have to start the whole replication process between
master and slave from scratch.
Hope my question is not confusing.
On Wed, Feb 17, 2010 at 10:31 AM, Greg Sabino Mullane <greg at endpoint.com>wrote:
> > I got the following error message and the bucardo process got
> > terminated. What happend was , a new column has been added to the
> > master database and the same column has not been added to slave.
> > Eventually, the replication was stopped with the following message.
> > Do we have to stop the replication process , incase if there is a DDL
> > change to master DB and make the same change to slave and resume the
> > bucardo_ctl? Please help, kind of serious issue for us.
> Yes. Postgres does not offer triggers on system tables, so there is
> no reliable way for Bucardo to detect such changes. The best way to
> make DDL changes is as you said: stop Bucardo, make the changes on both
> ends, then restart Bucardo.
> There are long-term plans to automate some of that and have Bucardo
> automatically detect and apply things like new columns to the slaves,
> but it's still a very basic proof of concept at this point.
> Greg Sabino Mullane greg at endpoint.com
> End Point Corporation
> PGP Key: 0x14964AC8
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Bucardo-general