On Tue, Nov 17, 2026 at 6:36 AM Zsolt Parragi <[email protected]> wrote: > > Hello > > I think the old cluster issue still exists eveen with v22/23, as > pointed out by claude: > > Set up a PG15 subscriber with an enabled subscription, with its table > in state i at clean shutdown, 50 extra databases (to make the check > phase longer), and pre_auth_delay=1 on the publisher to simulate a > slow connection. >
Thanks, I’ve reproduced both cases with a PG15 subscriber. The cause is that pg_upgrade starts the old cluster with -b, which only disables the logical replication launcher from PG17 onward. On older clusters, tablesync workers run while pg_upgrade is checking and dumping. v24 (attached) starts pre-17 old clusters with -c max_logical_replication_workers=0. With that, neither case reproduces from PG10 or PG15. Regards, Ajin Cherian Fujitsu Australia
v24-0002-Preserve-replication-origin-IDs-during-pg_upgrad.patch
Description: Binary data
v24-0001-Preserve-subscription-OIDs-during-pg_upgrade.patch
Description: Binary data
