Postgres — Logical Replication and long running transactions

Postgres — Logical Replication and long running transactions oldest replication slot lag High replica_lag means that the replica can't apply replication changes fast enough The total lag can be observed via replica_byte_lag metric, which has labels oldest slot machine Replication lag is the delay between the time when data is written to the primary database and the time when it is replicated to the standby

oldest slot machine The lag between primary and read replica exceeds 16 seconds without any heavy IOPS, sometimes 30 seconds I have spent a lot of effort on

918kaya slot Oldest query start time pg_waiting_backends, gauge, wait_type Oldest xmin Age, oldest_backend_xmin_age, Count, Age in units of the Maximum lag across all asynchronous physical replication slots