Re: bogus: logical replication rows/cols combinations

On 5/2/22 19:51, Alvaro Herrera wrote:
> On 2022-May-02, Tomas Vondra wrote:
> 
>> pgoutput.c is relies on relcache callbacks to get notified of changes.
>> See the stuff that touches replicate_valid and publications_valid. So
>> the walsender should notice the changes immediately.
> 
> Hmm, I suppose that makes any changes easy enough to detect.  We don't
> need a separate signalling mechanism.
> 
> But it does mean that the walsender needs to test the consistency of
> [rowfilter, column list, published actions] whenever they change for any
> of the current publications and it is working for more than one, and
> disconnect if the combination no longer complies with the rules.  By the
> next time the replica tries to connect, START_REPLICATION will throw the
> error.
> 
>> Why would we need to know publications replicated by other walsenders?
>> And what if the subscriber is not connected at the moment? In that case
>> there'll be no walsender.
> 
> Sure, if the replica is not connected then there's no issue -- as you
> say, that replica will fail at START_REPLICATION time.
> 

Right, I got confused a bit.

Anyway, I think the main challenge is defining what exactly we want to
check, in order to ensure "sensible" behavior, without preventing way
too many sensible use cases.

regards

-- 
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company