Horton

pg_createsubscriber --dry-run logging concerns

Peter Smith <smithpb2250@gmail.com>

View thread View on postgr.es
From: Peter Smith <smithpb2250@gmail.com>
To: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2025-10-01T02:29:56Z
Lists: pgsql-hackers 
Lately, I've been reviewing some pg_createsubscriber patches and have
been tricked by some of the logging.

The pg_createsubscriber has a '--dry-run' option/mode [1]

----------
--dry-run
Do everything except actually modifying the target directory.
----------

I've noticed that the logging in '--dry-run' mode is indistinguishable
from the logging of "normal" run, although functions like
create_publication(), drop_publication(), etc, are NOP in '--dry-run'
mode, because the actual command execution code is skipped. e.g.

if (!dry_run)
{
  res = PQexec(conn, str->data);
  ...
}

~~~

IMO, it's not good to fool people into thinking something has happened
when in fact nothing happened at all. I think the logging of this tool
should be much more explicit wrt when it is/isn't in dry-run mode.
Perhaps like this:

NORMAL
pg_log_info("creating publication \"%s\" in database \"%s\"", ...)

DRY-RUN
pg_log_info("[dry-run] would create publication \"%s\" in database \"%s\"", ...)

~~~

Thoughts?

======
[1] https://www.postgresql.org/docs/current/app-pgcreatesubscriber.html

Kind Regards,
Peter Smith.
Fujitsu Australia

Commits

  1. Split dry-run messages into primary and detail

  2. Log a note at program start when running in dry-run mode

  3. Use USECS_PER_SEC from datatype/timestamp.h

  4. Remove WaitPMResult enum in pg_createsubscriber

  5. pg_createsubscriber: reword dry-run log messages