cbbrowne at ca.afilias.info cbbrowne
Sun Jan 15 18:58:22 PST 2006
> On Fri, 2006-01-13 at 12:31 -0800, Alan Hodgson wrote:
>> On January 13, 2006 12:25 pm, Sven Willenberger <sven at dmv.com> wrote:
>> > So my question is whether there is any way to add tableB such that it
>> > now replicates from server2 to server1? Or is it possible to change
>> the
>> > subcription definition of server3 to set FORWARD = yes and could I
>> then
>> > set up replication from server3 to server1 as follows:
>> >
>> > server3 [origin] --> Set 3 [tableB] --> server1
>> >
>> > Or will I still fail due to the table_reloid constraint?
>>
>> You need to move the table to a third set that is replicated to the
>> necessary servers.
>>
>
> I came to the same conclusion about 5 minutes after I sent the email. I
> ended up dropping replication of the table and sequence, created a new
> set that had the one server as origin and the other 2 as subscribers,
> added the table and sequence to this new set and started subscribed the
> nodes. All worked well :-)

You could have handled this without dropping the table from replication...

Step 1.   Create the new set, set #3

Step 2.   It has no tables or sequences in it...

Step 3.   Subscribe set 3 as many times as needed to allow it to have
identical subscribers to set #1

Step 4.  Use MOVE SET to shift the table to set #3

Step 5.  Revise subscriptions for set #3

If the table is small, then this isn't material...




More information about the Slony1-general mailing list