Re: bogus: logical replication rows/cols combinations

On 5/7/22 07:36, Amit Kapila wrote:
> On Mon, May 2, 2022 at 6:11 PM Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> wrote:
>>
>> On 2022-May-02, Amit Kapila wrote:
>>
>>
>>> I think it is possible to expose a list of publications for each
>>> walsender as it is stored in each walsenders
>>> LogicalDecodingContext->output_plugin_private. AFAIK, each walsender
>>> can have one such LogicalDecodingContext and we can probably share it
>>> via shared memory?
>>
>> I guess we need to create a DSM each time a walsender opens a
>> connection, at START_REPLICATION time. Then ALTER PUBLICATION needs to
>> connect to all DSMs of all running walsenders and see if they are
>> reading from it. Is that what you have in mind? Alternatively, we
>> could have one DSM per publication with a PID array of all walsenders
>> that are sending it (each walsender needs to add its PID as it starts).
>> The latter might be better.
>>
>
> While thinking about using DSM here, I came across one of your commits
> f2f9fcb303 which seems to indicate that it is not a good idea to rely
> on it but I think you have changed dynamic shared memory to fixed
> shared memory usage because that was more suitable rather than DSM is
> not portable. Because I see a commit bcbd940806 where we have removed
> the 'none' option of dynamic_shared_memory_type. So, I think it should
> be okay to use DSM in this context. What do you think?
>

Why would any of this be needed?

ALTER PUBLICATION will invalidate the RelationSyncEntry entries in all
walsenders, no? So AFAICS it should be enough to enforce the limitations
in get_rel_sync_entry, which is necessary anyway because the subscriber
may not be connected when ALTER PUBLICATION gets executed.

regards

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