one more detail: pg_replication_slots.catalog_xmin shows the biggest age in the system.
How could transaction on catalogues may affect logical slot. Any clue here ?
Thank you,
Rado
On Wed, Jul 1, 2020 at 2:09 AM Radoslav Nedyalkov <rnedyalkov@xxxxxxxxx> wrote:
Hello All,we are in a situation where restart_lsn for logical replication slots is not advancing.We have two slots. Both are active, confirmed_flush_lsn also is updated.Client side all looks okay. remote_lsn for subscritions is advancing.The effect of restart_lsn being 'stopped' is the disk is filling up.Any idea what could be the cause?We will try to re-establish replication by dropping current slots , although we have to do it very carefully. We should avoid initial loading (these are 4T of data)Thanks and regards,Rado
