We have a test RDS postgresql 11.12 instance, configured for logical replication. The only thing connected to it is a 3rd party service, Hevo Data, which is managing sync to a separate Redshift instance. There is no incoming data - inserts, updates, deletes etc

Hevo does an historical load - a bunch of SELECTs basically, initially. This has finished; all data is present in the destination Redshift instance.

However, the problem is that the metric TransactionLogsDiskUsage continues to steadily rise. This would make sense if there was incoming data/new transactions and a replication subscriber wasn't fetching transaction logs, but I don't think either of those conditions are true.

What could be causing this metric to go up?


This is a duplicate of RDS Postgres oldest replication lag increases during inactivity even though Fivetran connector keeps synchronizing

Due to a lack of writes at the source, there's nothing to sync and (for whatever reason) postgresql retains write ahead logs in this case - until there is something to sync.

Doing an INSERT at the source caused TransactionLogsDiskUsage to drop to normal levels a few minutes later.

Not the answer you're looking for? Browse other questions tagged or ask your own question.