Re: Identify missing publications from publisher while create/alter subscription.
vignesh C <vignesh21@gmail.com>
From: vignesh C <vignesh21@gmail.com>
To: Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com>
Cc: Euler Taveira <euler@eulerto.com>, Dilip Kumar <dilipbalaut@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-05-03T14:29:15Z
Lists: pgsql-hackers
On Mon, May 3, 2021 at 11:11 AM Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> wrote: > > On Sun, May 2, 2021 at 10:04 PM vignesh C <vignesh21@gmail.com> wrote: > > > 5) Instead of adding a new file 021_validate_publications.pl for > > > tests, spawning a new test database which would make the overall > > > regression slower, can't we add with the existing database nodes in > > > 0001_rep_changes.pl? I would suggest adding the tests in there even if > > > the number of tests are many, I don't mind. > > > > 001_rep_changes.pl has the changes mainly for checking the replicated > > data. I did not find an appropriate file in the current tap tests, I > > preferred these tests to be in a separate file. Thoughts? > > If 001_rep_changes.pl is not the right place, how about adding them > into 007_ddl.pl? That file seems to be only for DDL changes, and since > the feature tests cases are for CREATE/ALTER SUBSCRIPTION, it's the > right place. I strongly feel that we don't need a new file for these > tests. > Modified. > Comment on the tests: > 1) Instead of "pub_doesnt_exist" name, how about "non_existent_pub" or > just pub_non_existent" or some other? > + "CREATE SUBSCRIPTION testsub2 CONNECTION '$publisher_connstr' > PUBLICATION pub_doesnt_exist WITH (VALIDATE_PUBLICATION = TRUE)" > The error message with this name looks a bit odd to me. > + /ERROR: publication "pub_doesnt_exist" does not exist in > the publisher/, Modified. Thanks for the comments, these comments are handle in the v7 patch posted in my earlier mail. Regards, Vignesh
Commits
-
Raise a WARNING for missing publications.
- 8f2e2bbf1453 15.0 landed