Why repmgr showing WAL replay is paused on all standby node?

Question

At repmgr cluster show command showing below error. Why it show WAL replay is paused on node2 & node3 as a warning. Here node2 and node3 is my standby node. Here I attached postgresql.auto.conf for node1 and node2. After PITR using Barman I found this issue. Why this is happen and how to solve it?

Answer 1

Since v12, the default recovery_target_action has been "pause". This lets you explore the system in read-only mode to make sure you are at the optimal point in time, before you fully open the database.

Since you didn't override that (by using barman's --target-action , for example), then "pause" is what you get.

You could fix this (without redoing the whole restore) by editing your conf files to add a recovery_target_action, or by just connecting to each one and invoking pg_wal_replay_resume().

Answer 2

I struggled with the same issue for almost 3-4 days and finally found that postgresql.auto.conf is an issue on standby. Please remove or comment lines in postgresql.auto.conf #restore_command = 'cp barman_wal/%f %p' #recovery_end_command = 'rm -fr barman_wal' #recovery_target_time = '2022-11-06 10:00:00'

I restarted my standby after cloning using repmgr and everything started to work. Hope this helps

Answer 3

PostgreSQL is unnable to replicate transaction that apply to the DDL part of SQL and when such a command is executed on the source database of the replication the streaming replication immediately stops.

Investigate that way to find which DDL command (CREATE, ALTER, DROP...) has benn executed on the source database...