On 29.03.21 11:13, Amit Kapila wrote:
This might or might not be valid for all logical replication solutions but in the publisher-subscriber model, it would easily lead to duplicate identifiers and block the replication. For example, when there are multiple subscriptions (say - 2) for multiple publications (again say-2), the two subscriptions are on Node-B and two publications are on Node-A. Say both publications are for different tables tab-1 and tab-2. Now, a prepared transaction involving operation on both tables will generate the same GID.
I think you are misunderstanding. This is about a globally unique identifier for a transaction, which has nothing to do with a GID used to prepare a transaction. This *needs* to be the same for what logical is the same transaction.
What GID a downsteam subscriber uses when receiving messages from some non-Postgres-provided output plugin clearly is out of scope for this documentation. The point is to highlight how the xid can be useful for filter_prepare. And that serves transaction identification purposes.
Regards Markus