https://www.postgresql.org/docs/12/logical-replication-restrictions.html

[PG Doc comments form]

The following documentation comment has been logged on the website:

Page: https://www.postgresql.org/docs/12/logical-replication-restrictions.html
Description:

Here, we should add one more point:

* Any column part of the replica identity needs to be of a type with a
equality operator.

Use Case:
When, I have added a table for logical replication(without any primary key)
which had a column with json datatype. I got below error at
target(subscription) db:

 LOG:  background worker "logical replication worker" (PID 17092) exited
with exit code 1

After that, I  altered the table at source db and converted the datatype to
jsonb instead of json. Then, I dropped and recreated the table schema with
jsonb column datatype on replica db and the error got resolved.

Re: pg_upgrade is not correct

[Bruce Momjian]

On Tue, Dec  8, 2020 at 03:13:30AM +, PG Doc comments form wrote:
> The following documentation comment has been logged on the website:
> 
> Page: https://www.postgresql.org/docs/10/pgupgrade.html
> Description:
> 
> If we use the following command to upgrade the standby server, then new
> cluster will create by old_cluster which is not correct.
> rsync --archive --delete --hard-links --size-only --no-inc-recursive
> old_cluster new_cluster remote_dir
> I used following command and this one is correct:
> rsync --archive --delete --hard-links --size-only --no-inc-recursive
> new_cluster old_cluster remote_dir 
> So the only thing need to change is the place of old_cluster and new_cluster
> in rsync command.

I am not sure how to reply.  These instructions have been unchanged for
a few years.  Why is the order you suggest correct?

-- 
  Bruce Momjian  https://momjian.us
  EnterpriseDB https://enterprisedb.com

  The usefulness of a cup is in its emptiness, Bruce Lee