Re: Implicit table removal from logical replication publication - Mailing list pgsql-general

From Cory Nemelka
Subject Re: Implicit table removal from logical replication publication
Date
Msg-id CAMe5Gn2RKQVWajxNSjDeCGP=SE1HYSYZsNcvHun=a7CPGUKnNg@mail.gmail.com
Whole thread Raw
In response to Re: Implicit table removal from logical replication publication  (Vijaykumar Jain <vijaykumarjain.github@gmail.com>)
Responses Re: Implicit table removal from logical replication publication  (Vijaykumar Jain <vijaykumarjain.github@gmail.com>)
List pgsql-general


On Thu, Jun 10, 2021 at 12:39 PM Vijaykumar Jain <vijaykumarjain.github@gmail.com> wrote:
Wow, the drop table silently removes entry from publication without any logs.

I could not find any monitoring view to help me figure out if the publication is broken due to ddl change.
pg_stat_replication on publisher, and pg_stat_subscription on subscriber only help with lsn based lag.
unless this is not an issue but a feature ?
i'll check more on the better part of monitoring logical replication stuff.

but for your case, 
you can set up an event trigger that would avoid dropping the table. 
basically any drop of a table or any ddl that would break publication.


you can have a custom query filter that would prevent dropping of objects part of publication accidentally.

and then you want to exclusively drop the table, once not part of publication, you have to first remove the table from publication and then drop.

I have not run this in production, so I believe others may chime in, but logical replication issues from logs are not the best.
I am happy to be corrected.
I'll update on more scenarios.

is pg_publication_tables what you are looking for?


--
--cnemelka

pgsql-general by date:

Previous
From: Vijaykumar Jain
Date:
Subject: Re: Implicit table removal from logical replication publication
Next
From: Vijaykumar Jain
Date:
Subject: Re: Implicit table removal from logical replication publication