Dear Amit, > Sorry, my analysis was not complete. On looking closely, I think the > reason is that we are allowed to upgrade the slot iff there is no > pending WAL to be processed.
Yes, the guard will strongly protect from data loss, but I do not take care in the test. > The test first disables the subscription > to avoid unnecessary LOGs on the subscriber and then stops the > publisher node. Right. Unnecessary ERROR would be appeared if we do not disable. > It is quite possible that just before the shutdown of > the server, autovacuum generates some WAL record that needs to be > processed, Yeah, pg_upgrade does not ensure that autovacuum is not running *before* the upgrade. > so you propose just disabling the autovacuum for this test. Absolutely correct. Best Regards, Hayato Kuroda FUJITSU LIMITED