Ensure host B is your active host in the dual master
Rest of instructions are on host A
As reading from A.tablename will probably crash it, try this with another tablename to be familiar with what its doing. pt-table-sync --dry-run --print --execute --sync-to-master h=hostB,D=db,t=sometablename
look closely at the above.
SET GLOBAL innodb_file_per_table=1 (...
Seconds_behind_master doesn't exactly measure the replication lag. It measures the difference between the timestamp of the last binlog event fetched by the IO thread, versus the timestamp of the last binlog event executed by the SQL thread.
This can become misleading for example if the master has more binlogs that the replica hasn't fetched yet. If the ...
Replication Slots filling up disk?
Even if you specify use_slots in Patroni's configuration, the slot will be automatically removed if the replica fails to renew it's member key in the DCS (distributed consensus store). By default this means a slot for a disappeared replica will be dropped after 30 seconds (this corresponds to the ttl setting, which ...
Answering this from the future - when I have seen this, it means there have not been any writes to the primary (upstream) database in a while.
If there are no recent transactions on the primary, there are no recent transactions to replay on the follower, so pg_last_xact_replay_timestamp() will just report the time of the last write to primary.