Re: Seems that majordomo stopped working forpgAdmin-support?

On Wed, 2007-09-19 at 09:20 -0700, Daniel B. Thurman wrote:
>
> I send an elaborate configuration image of the pgAdmin setup with all
> the properties show for both the Master and Slave. I was hoping that
> you could see what I had done as far as setting things up. I was also
> hoping that this might aide others who may be trying to see the setup
> for what it should/might be? Perhaps this is over-the-top but I am
> trying to figure out how to setup the replication implementation.
>
> So far, I did get the (manual) slons working but as posted as a follow
> up, I am having all sorts of issues now.
>
> The problems I have encountered:
>
> 1) I was unable to run slonik with the configuration file that had the
> Cluster Name = Copper-Master-Cluster entry because I *think* that
> slonik did not like '-'. So I ended up terminating the manually
> running slons, deleted the Replication for the master side, and
> attempted to do the same for the slave side, and I discovered that
> I was not allowed to do that because there was an error message saying
> that I was not allowed to delete the replication sets because there
> was a dependency issue that being it was a provider?
>
> So I went the whole hog trying to find every entry in the replication
> structure and tried to delete the listeners, the nodes, etc. and again
> was not able to because of another dependency (the replication sets),
> so it turned out to be circular. Giving up, I ended up deleting the
> entire slave-side database, rebuilt it, and then was able to define a
> new Cluster Name but without the '-' in the name.

Whic are you using, pgAdmin or Slonik? Do not try to use both while you
are learning - it will lead to inevitable pain and suffering.

> 2) Now at this point, I am still having a problem trying to figure out why
> I am unable to get replication to work with the slons running. I see no
> errors reported from the debug strings on both master/slave sides and the
> odd thing here is that, I tried to update an entry in the Master-side database
> for the MyTest.cars table and I get an error reporting: "An error has occurred:
> ERROR: Slony-I: Table cars is replicated and cannot be modified on a subscriber
> node"...

Sounds like you are trying to update data on the slave, not the master.

> The odd thing here (maybe?) that I was previously unable to define a new sequence
> because there was nothing available in the Sequence dropdown list, so I ignored
> this and proceeded to create a subscrition anyway.
> Again, once the subscription is created, I cannot 'start over' and delete the entire
> replication set both for the master and slave sides.

If you need to start over, kill the slons, drop the slave database, and
remove the _<clustername> schema from the master iirc.

Try a very simple setup first, using just a couple of tables with a
master and a slave. You might find Rob's blog useful:
http://people.planetpostgresql.org/xzilla/index.php?/archives/200-Alpha-testing-Slony-on-win32-Crib-Notes.html

Regards, Dave

Regards, Dave