There are a lot of threads here about initial loads being slow. There are some ways to work around the slowness by manually loading the tables and setting state but nothing official. I know there's some work being done by Meltano for "batch mode" or something like that but I don't know where that is in terms of actually being usable.

Conceptually the steps are: 1. Capture the "binlog position" for the table in question ("binlog position" is a MySQL thing - I'm not sure what the PG equivalent of "binlog position" is, unfortunately, but you can look at the code for the tap to see what it does) 2. Dump the data from the source DB using that DB's tools (psql I think in your case, or maybe pg_dump?) 3. Import the data into the destination DB using that DB's tools 4. Manually set the state of the table in meltano's internal DB a. Create a state.json that specifies the "binlog position" for the table in question (you can run

meltano state get some_id

to see what the state data is supposed to look like b. Run

meltano state set some_id --input-file state.json

to load the state for the table into meltano's internal DB From there, meltano will read changes to the table from the source and push them to the destination as if it had loaded the table itself from the beginning.

Actually between steps 2 and 3 you have to create the table too 😉 To make all this a little smoother you can create a copy of the table in question that contains only one row, let meltano replicate that normally, then a) grab the state data for that as a template for the real table, and b) grab the CREATE TABLE for that to create the real table.

It's an unfortunate problem to have to wrestle with when first learning the tool. 😞 Working through all that will teach you a lot about how meltano works though, on the bright side...