On Wed, Aug 11, 2021 at 4:57 PM Dave Cramer <davecramer@gmail.com> wrote: > > On Wed, 11 Aug 2021 at 07:24, Amit Kapila <amit.kapila16@gmail.com> wrote: >> >> On Wed, Aug 11, 2021 at 1:18 AM Dave Cramer <davecramer@gmail.com> wrote: >> > >> > Reviving this thread >> > >> > 2021-08-10 19:05:09.096 UTC [3738] LOG: logical replication apply worker for subscription "sub_mycluster_alltables" has started >> > 2021-08-10 19:05:09.107 UTC [3739] LOG: logical replication table synchronization worker for subscription "sub_mycluster_alltables", table "t_random" has started >> > 2021-08-10 19:05:12.222 UTC [3739] LOG: logical replication table synchronization worker for subscription "sub_mycluster_alltables", table "t_random" has finished >> > 2021-08-10 19:05:14.806 UTC [3738] ERROR: could not receive data from WAL stream: ERROR: publication "sub_mycluster_alltables" does not exist >> > CONTEXT: slot "sub_mycluster_alltables", output plugin "pgoutput", in the change callback, associated LSN 0/4015DF0 >> > 2021-08-10 19:05:14.811 UTC [175] LOG: background worker "logical replication worker" (PID 3738) exited with exit code 1 >> > >> > >> > select * from pg_publication; >> > -[ RECORD 1 ]+------------------------ >> > oid | 16415 >> > pubname | sub_mycluster_alltables >> > pubowner | 10 >> > puballtables | t >> > pubinsert | t >> > pubupdate | t >> > pubdelete | t >> > pubtruncate | t >> > >> >> By any chance, did you dropped and recreated this publication as >> mentioned in your first email? If so, I think this can happen because >> of our use of historical snapshots to consult system catalogs. > > > In this case, no. > > I am suspecting this error comes from pgoutput though. >
I think it is and the context is generated via output_plugin_error_callback. Is this reproducible for you and if so, can you share a test case or some steps to reproduce this? Does this work and suddenly start giving errors or it happens the very first time you tried to set up publication/subscription? I think some more details are required about your setup and steps to analyze this problem. You might want to check publication-side logs but not sure if get any better clue there.
In this case I am the messenger. I will try to get a repeatable test case.
