On Mon, Dec 18, 2017 at 10:53 AM, Aleksander Alekseev
<a.alekseev@postgrespro.ru> wrote:
> Hello hackers,
>
> Thanks you a lot for your feedback. I modified the project description
> to make it more clear that it implies augmenting an existing HA
> solution, particularly Stolon, and doesn't imply solving existing
> limitations of logical replication like lack of DDL replication.
>
> Removing some of these limitations or adding logical replication support
> to other existing HA applications like RepMgr or Patroni or corosync /
> Pacemaker or maybe Postgres-XL sounds like good ideas for other GSoC
> projects. I strongly encourage to add these ideas to wiki anyone who is
> willing to be a mentor.
>
> Unfortunately I don't have any experience of using these applications
> and have no idea how they behave in different corner cases like
> netsplit, slow network, etc. Also I don't have much motivation to figure
> this out because I'm pretty happy with Stolon. For these reasons I
> personally can't be a mentor for corresponding projects.
>
> Also today I'm going to contact Simone Gotti, the main developer of
> Stolon, to let him know about this thread. I wonder what are his
> thoughts on all this.
Hi,
here at SorintLab we'll be very happy to provide help and mentor this project.
I read and agree with your concerns since some missing features in
logical replication cannot provide, at the current stage, a complete
seamless HA experience like with streaming replication.
But I believe that enhancing stolon to use logical replication will be
a really useful scenario where we can test the features of logical
replication for HA and use it also for future enhancements (like major
upgrades with near zero downtime).
At the same time this could also push me and other contributors to
implement the missing logical replication features needed to achieve
seamless HA.
Cheers,
Simone.
>
> --
> Best regards,
> Aleksander Alekseev