Horton

Re: Handle infinite recursion in logical replication setup

vignesh C <vignesh21@gmail.com>

View thread View on postgr.es
From: vignesh C <vignesh21@gmail.com>
To: Amit Kapila <amit.kapila16@gmail.com>
Cc: "shiy.fnst@fujitsu.com" <shiy.fnst@fujitsu.com>, Peter Smith <smithpb2250@gmail.com>, "houzj.fnst@fujitsu.com" <houzj.fnst@fujitsu.com>, Dilip Kumar <dilipbalaut@gmail.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, "kuroda.hayato@fujitsu.com" <kuroda.hayato@fujitsu.com>, "Jonathan S. Katz" <jkatz@postgresql.org>
Date: 2022-09-08T03:27:00Z
Lists: pgsql-hackers 
On Thu, 8 Sept 2022 at 08:23, Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> On Thu, Sep 8, 2022 at 7:57 AM vignesh C <vignesh21@gmail.com> wrote:
> >
> > There is a buildfarm failure on mylodon at [1] because of the new test
> > added. I will analyse and share the findings for the same.
> >
> > [1] - https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=mylodon&dt=2022-09-08%2001%3A40%3A27
> >
>
> The log is as below:
>
> error running SQL: 'psql:<stdin>:1: ERROR:  could not drop relation
> mapping for subscription "tap_sub_a_b_2"
> DETAIL:  Table synchronization for relation "tab_new" is in progress
> and is in state "s".
> HINT:  Use ALTER SUBSCRIPTION ... ENABLE to enable subscription if not
> already enabled or use DROP SUBSCRIPTION ... to drop the
> subscription.'
> while running 'psql -XAtq -d port=50352 host=/tmp/WMoRd6ngw2
> dbname='postgres' -f - -v ON_ERROR_STOP=1' with sql 'DROP TABLE
> tab_new' at /mnt/resource/bf/build/mylodon/HEAD/pgsql.build/../pgsql/src/test/perl/PostgreSQL/Test/Cluster.pm
> line 1860.
> ### Stopping node "node_A" using mode immediate
>
> This clearly indicates the problem. We can't drop the relation till it
> is marked ready in pg_subscription_rel and prior to dropping, the test
> just does $node_A->wait_for_subscription_sync($node_B, $subname_AB2);.
> This doesn't ensure that relation is in 'ready' state as it will
> finish even when the relation is in 'syncdone' state.

I agree with the analysis, adding wait for ready state before dropping
the table approach looks good to me. I will prepare a patch for the
same and share it.

Regards,
Vignesh

Commits

  1. Improved CREATE SUBSCRIPTION message for clarity

  2. Fix the test case introduced by commit 8756930190.

  3. Raise a warning if there is a possibility of data from multiple origins.

  4. Add wait_for_subscription_sync for TAP tests.

  5. Bump catversion for commit d8cd0c6c95c0120168df93aae095df4e0682a08a.

  6. Allow users to skip logical replication of data having origin.

  7. Improve two comments related to a boolean DefElem's value

  8. doc: Fix man page whitespace issues