Re: How to Handle Sequences in Logical Bi-Directional Replication

Hi PGSQL DBA,

Here are my 2 cents on this

- Using different ranges for node1 and node2 ( Odd, Even)
- Concatenating node_id to sequence number. ( myseq.nextval || node_id).
- UUID , though there won't be any risk of conflict/collusion but will
take more storage.

Regards,
Muhammad Ikram
Bitnine global.

On Thu, Jun 13, 2024 at 9:13 AM Matti Linnanvuori <
mattilinnanvuori(at)protonmail(dot)com> wrote:

> Hi!
>
> I have planned replacing sequences with UUIDs, which work in distributed
> networks.
> On Thursday, June 13th, 2024 at 04.23, PGSQL DBA <pgsqldba(dot)1987(at)gmail(dot)com>
> wrote:
>
> Hi Experts,
>
> I am reaching out to seek your guidance and expertise on handling
> sequences in a bi-directional logical replication setup.
>
> I am planning to set up two PostgreSQL-16 database servers with logical
> bi-directional replication, using the *`origin=none`* configuration. In
> this setup, my application will be connecting to either of the database
> servers for DML operations.
>
> Given this scenario, I am looking for the best practices to handle
> sequence values to ensure consistency and avoid conflicts. Specifically, I
> would like to understand:
>
> 1. How to configure sequences to avoid duplicate values in both database
> servers.
> 2. Whether there are specific settings or configurations in PostgreSQL
> that facilitate the management of sequences in a bi-directional replication
> environment.
> 3. Any recommendations or strategies to ensure smooth and consistent
> sequence management across the two servers.
>
> I appreciate any guidance you can provide on this matter.
>
> Thank you!
>
>
>

--
Muhammad Ikram